Owl meets beehive: how impact assessment and governance relate

Abstract

This article analyses the complex relationship between impact assessments (IAs) of all types (such as comprehensive, regulatory, economic, social or environmental IAs) and their governance environment, using an analytical framework based on the concepts of governance styles and metagovernance. It is argued that each governance system builds on specific values, traditions and history and produces specific mixtures of hierarchical, network and market styles of governance (with or without an explicit metagovernance approach). Although governance can be considered as a non-normative perspective on polity and politics, the normative dimension of governance practice results in, for example, conflicting convictions about which type of knowledge or ‘evidence’ is valid for IA processes. This is particularly relevant because IAs have an important role in improving the knowledge base of governance. The concrete governance system in a specific country or other administrative entity influences the design and governance of IA systems positively or negatively, which leads to a variety of challenges. The conclusion is drawn that understanding the governance context and its dynamics can help improving IA governance.

Keywords::

• impact assessment
• governance
• hierarchical governance
• network governance
• market governance
• metagovernance
• EIA
• SEA

1. Introduction

Impact assessment (IA) is a forward-looking instrument that seeks to advise decision-makers proactively on the potential advantages and disadvantages of a proposed action (Partidário 2012). There are comprehensive IA procedures, such as the one used by the European Commission – the executive administration of the European Union (EU) – which covers regulatory, economic, social and environmental impacts, and IAs which focus on only one or two of these types of impacts. IA systems may target strategies and policies, legislation, plans, programmes or projects, sometimes even beyond the legal requirements; although the EU Strategic Environmental Assessment (SEA) Directive only mentions plans and programmes, in some European countries SEA is also applied to strategies and policies, for example in England (Therivel & Fischer 2012). The wider use of SEA is also advocated by the World Bank, which recognises SEA as ‘a key means of integrating environmental and social considerations into policies, plans and programs, particularly in sector decision-making and reform’ (World Bank 2013).

The term governance emerged when it became recognised that government is not the only actor dealing with societal problem-solving. In this article, I use a broad governance definition, which covers not only the processes of decision-making, but also the related institutions, instruments and the roles of non-governmental actors. The term will be introduced more elaborately below (Section 2). IA is widely considered as an important instrument of governance. This implies that governance systems determine the conditions under which IA takes place, including who is involved in running it and participating in it, as well as the values behind it. Therefore, the success of IA practitioners' work, including the governance of the IA procedures they manage (IA governance [IAG]), is influenced by the design and dynamics and by the values and traditions of the governance environment in which IA takes place. I will argue that the links between the concepts of IA and governance account for many conflicts, for example with regard to the timing and intensity of public consultation and stakeholder participation. Most of these conflicts are well known and described elsewhere already but seldom explicitly in the context of the relationships between IAs and their governance environment. It will be shown that not addressing such problems in their context can undermine the efficacy of both IA and governance systems.

One of the reasons why the governance dimension of IA has only recently started receiving substantial attention is that in the early years of, for instance, Environmental Impact Assessment (EIA, introduced in the 1980s), a majority of academic literature approached IA as an information-processing model. EIA scholars were biologists, planners, engineers and lawyers, who believed that EIA was a technical process, governed by scientific and technical rationality, comprising the collection of relevant technical information, and resulting in decisions on, ideally, technocratic merits rather than any political considerations (Bartlett & Kurian 1999). This preference for rationality is still there, but experience has grown among IA practitioners and theorists that the complexity of problems and the difficult logics of human behaviour including power dynamics (see e.g. Jiliberto 2012; Cashmore & Axelsson 2013; Cashmore & Richardson 2013; Hansen et al. 2013; Partidário & Sheate 2013) are phenomena that should also be taken into account. A similar wider view has been taken up within political science and economic theory. The rather rational approach to IAs has led to the situation that until recently most research on IA systems did not fully take into account the complexity, constraints and opportunities of the governance environment. IA tended to be examined in isolation from broader trajectories of public management reform and the administrative context (Radaelli et al. 2010). SEA literature was considered to be insular and not cognisant of the policy-making process which it intended to influence (Nitz & Brown 2001). Gradually, more attention is being paid to governance issues such as the role of knowledge in strategic contexts (e.g. Van Buuren & Nooteboom 2010; Richardson & Cashmore 2011; Partidário & Sheate 2013). Social impact assessment became defined as ‘the process of managing the social issues of development’ (Esteves et al. 2012), which is close to how social governance could be defined. Another important development supporting IA practice is the fact that most international financial institutions (IFIs) meanwhile include basic governance requirements in their lending conditions, and that intergovernmental organisations like UNDP promote optimising governance through their programmes.

The challenge I discuss in this article can be expressed in a metaphor of the owl and the beehive. In real life, owls and bees have nothing more in common than their ability to fly. However, on the meta-level of symbolism, the owl's knowledge, wisdom and ‘intelligence’1 (IA) and the beehive's organisation qualities, sense of order, industry, cooperation and hard work (governance) can form a winning team. The issue is how to connect these qualities in a productive way. With an analytical framework based on the concepts of governance styles and of metagovernance (elaborated in Section 2), four questions will be addressed which constitute important dimensions of the relationship between IA and governance (Figure 1):

• What are the opportunities and constraints of different governance systems for IA practice? (Section 3.1)

  Figure 1 Impact assessment and governance: four questions.

  

• How may understanding of governance systems help to explain commonly reported IA problems/deficiencies? (Section 3.2)

• How may IA contribute to the goals of different governance styles and their composite systems? (Section 3.3)

• What does understanding the governance environment imply for improving the governance of IAs? (Section 4).

The examples from IA theory and practice in this article are to a greater extent taken from comprehensive IA and SEA than from more sectoral IAs (e.g. EIA). This is for the reason that more literature on the former types has described tensions between IA and governance.

2. Conceptual framework: the relationship between IA and governance

In the Introduction, a broad definition of IA was presented. Since I would like to allow some complexity in the exploration of the relationship between IA and its governance environment, a broad definition of governance is needed as well, which covers the whole variety of governance practices. Therefore, in this article governance will be considered in a non-normative way, as addressing how decisions are made, problems are tackled and opportunities are created. Governance is more than ‘what government does’ as it includes what other actors do. Governance may have many forms, ranging from forcing to nudging, co-creation or enabling, depending on the circumstances and the objective of the ‘governor’. Governance therefore includes polity (the institutions and instruments) and politics (the processes). It addresses cross-cutting issues like the choice of institutions (i.e. the rules of the game, including laws and formal or informal organisations), instruments and processes, as well as decisions about the roles of those who will be affected (Mayntz 2004). In addition, the term governance covers the relations with those who are ‘governed’, regardless of whether they are considered as subordinates, partners or clients. Therefore, governance is a relational concept.

To help understanding how governance works, scholars have identified three main styles of governance (e.g. Thompson et al. 1991; Meuleman 2008): hierarchical, market-driven and network-oriented governance. These styles should be considered as ‘ideal types’ (Weber 1952): logical constructions which can be used to aid in the understanding of reality. This is useful because the three styles usually occur in mixed forms and seldom in a ‘pure’ form. The following broad definition (after Meuleman 2008) covers all three basic styles: ‘Governance is the totality of interactions in which government, other public bodies, private sector and civil society participate (in one way or another), aimed at solving public challenges or creating public opportunities’. This definition implies that governance as a concept is not per se normative: it is only normative when applied in a specific case or when a specific governance style is chosen as a recipe for all problems. In practice, most governance definitions are normative because they (implicitly) show a preference for a specific set of values. ‘Good governance’ as defined by the World Bank (1994), for example, for years focused on hierarchical values like legitimacy, effectiveness and control and on market values like efficiency and decentralisation. Other concepts such as network governance and related concepts such as reflexive governance and adaptive governance implicitly build on the assumption that actors value empathy, trust, cooperation and consensus more than, for example, legitimacy, authority, coercion or efficiency. Both the ‘market’ and ‘network’ schools of governance are to a certain extent anti-state and/or against hierarchical governance: governance is ‘a system of rule that works only if it is accepted by the majority (or, at least, by the most powerful of those it affects), whereas governments can function even in the face of widespread opposition to their policies’ (Czempiel & Rosenau 1992). Many theorists, however, consider also hierarchical governance to be an essential style in the governance mix (Hill & Lynn 2005; Olsen 2006; Meuleman 2008). The normative dimension of governance is prominent in the definition by In't Veld (2013): ‘Governance is a collection of normative insights (“how should …”) about the organisation of influence, steering, power, checks & balances in societies’.

The logical implication of the relational and normative dimensions of governance is that there is no preset governance approach for any particular problem. Each case has to be tailored to the statutory and informal frameworks in which it occurs. The importance of the context is also acknowledged by the World Bank: ‘Supporting open and collaborative governance will enable local change agents to achieve development results in their own contexts’.2 Which governance style dominates in a specific situation is influenced by the context. Some political-administrative traditions tend towards a legislative approach for societal problem-solving, while others favour efficiency as the key driver; still others promote a consensual approach.

In governance theory, the question how to design and manage situationally suitable combinations of different approaches is addressed by the concept of metagovernance (see e.g. Jessop 1997; Meuleman 2008). Metagovernance is a key challenge, because, as Davis and Rhodes (2000) argued:

… the future will not lie with markets, or hierarchies or networks but with all three and the trick will not be to manage contracts or steer networks but to mix the three systems effectively when they conflict with and undermine one another.

Metagovernance is non-normative when no a priori preference for one style in the governance mixture exists, but normative when such a preference exists, e.g. for network governance. The latter is called first-order, the first second-order metagovernance (Meuleman 2011).

Each governance environment (the whole of rules and laws, institutional setting, policy instruments, division of tasks and roles of governments and societal stakeholders, within a particular country or other boundary or transboundary units) affects the governance of IAs. A government-centred, hierarchical governance system has much strength but may at the same time lack the ability to deal with transparency, consultation and participation, which are characteristics often seen as essential for IA. The legal dimension and other aspects of governance such as the form of decision-making lead to considerable variation in the practice of sustainability IAs3 (Bond et al. 2012). Another important dimension is the perceived state of the economy. The role or scope of IA, in general, might be interpreted differently by governments facing economic recession than by governments who are in ‘sustained’ economic growth (Bond et al. 2012). Economic recovery governance on the level of the EU (the ‘European Semester’) can be a driver for strengthening comprehensive IA systems across nations (see Section 3.1).

Figure 2 summarises the above in an analytical framework for this article: the three governance styles and the concept of metagovernance applied to the relationship between IA and governance.

Figure 2 Impact assessment, governance and metagovernance: analytical framework (own composition).

From the above introduction of the terms IA (as forward-looking instrument) and governance (as a relational and normative approach to getting to decisions and implementation), it can be concluded that it is important to think about the wider governance environment in which IA processes are carried out before determining which specific governance approach would be suitable for a concrete IA procedure, or in other words how IAG should look like. However, before I discuss how IAG might be improved (Section 4), Section 3 addresses underlying questions.

3. Analysis

3.1. Opportunities and constraints of governance systems for improving IA practice

3.1.1. Governance programmes

The first question is about how governance systems influence IA practice. Each IA system is embedded in a certain governance environment, which therefore also determines the conditions under which IA takes place. An example of how a governance system can stimulate the use of comprehensive IAs4 is the annual governance cycle on implementation of the Europe 2020 Strategy on ‘smart, sustainable and inclusive growth’, the ‘European Semester’.5 The European Commission proposes every year ‘Country Specific Recommendations’ (CSRs) which are then reviewed by the 28 Member States. The Semester is, on the one hand, a formal governance process (the European Commission proposes Recommendations, the national Ministers debate the merits, the Heads of States and Governments decide). On the other hand, it is also a voluntary process, i.e. non-compliance is not (directly) followed by sanctions. The Semester process contains, since its start in 2011, in all its deliverables6 a chapter on Modernisation of Public Administration, in which IA may figure. Most EU Member States apply some sort of (regulatory or broader) IA for the policies, strategies or legislation to address the Semester CSRs. A recent legal requirement supports this: the 7th Environment Action Programme of the EU (European Union 2013) states that measures taken by the EU and its Member States supporting a move towards a ‘circular economy’ should be ‘in line with the principles of smart regulation and, where appropriate, subject to a comprehensive IA’.7

There is a growing understanding that economic and social reform measures in the context of the Semester should undergo comprehensive IA. For example, the 2014 Semester Staff Working Document (SWD) for Luxembourg states that ‘comprehensive impact assessment (…) could provide the costs and benefits of different policy options, and also support precautionary action by estimating the costs of inaction’. It explains that

from case studies, it is clear that the costs of inaction have often been grossly underestimated in the past.8 Luxembourg could save costs (e.g. costs of inaction, lock-in effects, indirect and long-term effects) by introducing environmental integration criteria in its present system of impact assessment. (European Commission 2014)

Similar statements appear in the 2014 SWDs for Austria, Belgium, Italy and the Netherlands. A first conclusion could be that economic and social governance in times of crisis recovery and financial austerity can offer an opportunity for improving comprehensive IA systems, because the call for ‘evidence-based policy’ and understanding positive and negative impacts of policy and legal proposals may be higher than when the economy does not pose complex problems. The degree to which societal stakeholders are involved in such IAs varies widely across the different Member States with their respective administrative governance cultures.

Governance programmes can also put in place basic requirements of IA systems, beyond stimulating informed decision-making, in situations where there is no comprehensive or e.g. environmental IA legislation. The UNDP/UNEP Poverty and Environment Initiative is an example of a programme that invests heavily in capacity building and involvement of civil society and local organisations in less developed countries with a centralist, hierarchical governance tradition. Such a programme can create the demand for further involvement, including in strategic decision-making. In Tajikistan this already contributed to the establishment of SEA legislation (Challe et al. 2014). It would be worthwhile to investigate whether this effect has occurred more widely.

3.1.2. Administrative cultures and political settings

Administrative cultures and political settings belong to the factors which influence IA practice (Jacob et al. 2008; Lobos & Partidário 2014). They are relevant as success or failure of IA relates to its compatibility with the preferred governance style of the decision-makers in a specific case (e.g. Monteiro et al. 2014, who compare SEA practice in China, Denmark, Vietnam and the Netherlands). The dominant culture within one organisation/administration or country, region or even city co-determines which governance approach will work (Meuleman 2013a). National cultures may show an underlying ‘default’ governance approach (Meuleman 2008), which is – broadly speaking – more market-driven in Anglo-Saxon countries9 with their tradition of liberal economics, more network-driven in northwest Europe and more hierarchical in central, eastern and southern Europe and in many Asian and South-American countries.

Administrative cultures influence not only which type of solution is preferred in the policy options assessed in IA, but also how ‘evidence’ is defined. Successful IA experiences – measured against their objectives – from elsewhere could therefore better be called ‘good practices’ than ‘best practices’, as the latter term suggests universal applicability: transferability of different IA models from one jurisdiction to another has shown to be difficult (Jacob et al. 2008). Such place-based determinants seem important for IA processes. Learning from international experience should therefore not focus on copying (IA) structures or institutions but on creating the conditions for mechanisms to operate (Radaelli et al. 2010).

A case study in Madagascar may serve as an example of the importance of compatibility between IA and governance. A combined set of criteria for good community governance and sustainability assessment, adapted to the context of the case study, provided a useful tool in pinpointing strengths and weaknesses, and shed light on sustainability issues and possible improvements (Vincent & Morrison-Saunders 2013). The principles for good community governance which the authors selected were of a network governance (‘collaboration, fairness and equity’) and hierarchical governance (‘accountability’) nature, which seems compatible with the core idea of ‘community governance’. A last example of the impact of administrative cultures and settings is observed in the European Commission. Its comprehensive IA system is coordinated by the Secretariat-General, which is close to the Commission President and hence close to the political power centre. It is therefore not surprising that the one unifying narrative theme found in an in-depth linguistic-narrative analysis of a sample of eight Commission IA procedures is ‘coordination and harmonization’ (Radaelli et al. 2013), which are principles of hierarchical governance.

The favoured governance styles within politico-administrative systems will influence any review or revision of an existing legal IA system. In line with the political promise of the its White Paper on Governance (European Commission 2001), the EU attempts to combine hierarchical governance with the two ‘new modes of governance’ (Héritier 2002), network and market governance. The recent revision of the EU EIA Directive (European Union 2014) is a good illustration. It has a strong market governance footprint: flexibility, cost reduction, streamlining and administrative simplification are expressed, e.g., in Articles on a ‘one stop shop’ and on a relatively modest reporting requirement (efficiency). Elements of hierarchical governance can also be found: several new provisions are about ensuring transparency, and also the new requirement to use ‘competent experts’ and ‘sufficient expertise’ points at a hierarchical view on what is good knowledge (namely ‘authoritative’ knowledge). Finally, network governance ideas were served by the inclusion of better public consultation requirements, ‘prompt’ information requirements of decisions and the acknowledgement of multilevel governance (involvement of local and regional governments). Further research could reveal to which extent the governance mixture supporting the revised EIA Directive was the result of a conscious metagovernance approach by or between the main actors – Commission, Council and Parliament. Another example is the revision of the federal Canadian EIA system in 2012, which was strongly influenced by the market governance trend of simplification: it reduced the scope of EIAs, shortened timelines for decisions, will have fewer agencies and federal departments involved in the EIA process, and fewer persons or groups that may have an ‘interested party’ status (Ingelson 2013).

3.2. Commonly reported IA problems explained from a governance perspective

The next question is a sub-question of the first: how may the characteristics of the ideal-typical governance models lead to tensions with particular IA good practice norms, and how may an understanding of governance systems help explaining commonly reported IA problems/deficiencies?

3.2.1. Typical weaknesses of governance styles

Some of the challenges experienced by IA practitioners can be traced back to inherent weaknesses or mutually undermining characteristics of specific governance styles (Jessop 2003; Meuleman 2008). Hierarchical governance fosters stability (but also its usually more negatively perceived form rigidity), predictability (but also low capacity to react to unexpected events), rule of law (but also abuse of power) and clear lines of command (but also much bureaucracy and the resulting delays). Network governance could lead to more inclusive IAs, but includes the risk of losing focus (‘we should talk to everybody’) and resulting in a consensus-building process in its own right. Market governance principles for decision-making can lead to overestimating the relevance of price and techno-commercial efficiency. This style of governance may prioritise efficiency above the original objective of the policy, plan or project. Tangible and non-tangible effects may become monetised to the detriment of a more qualitative/narrative approach that would – if included – make the IA process richer and with more options available. A strong emphasis on monetisation tends to under-represent qualitative information and quantitative data that cannot be ‘monetised’. It has been argued that the precautionary principle, prominent in IA theory and practice, is under pressure from the ‘better regulation’ agenda of the New Public Management movement (Meuwese 2008), which focuses on cost reduction, monetisation and deregulation and is therefore an example of market governance. Such pressure stimulates IA practitioners to monetise all types of impacts (ecosystem services, natural accounting, health damages, etc.).

A governance system that is not compatible with the type of problem for which the IA process is designed may invoke problems. For example, hierarchical and market-type mechanisms may be unsuitable when the IA subject is contested and complex. On the other hand, network governance may be too slow for routine issues and too indecisive for disaster prevention projects. Routine issues profit from a market governance attitude (efficiency, competition, incentives). An example of how governance styles may undermine each other is that co-operative behaviour, essential in network governance, ‘can collapse under the impact of competition in whatever form’ (Davis & Rhodes 2000). Competition is typical for market governance. If we were able to connect observed IA challenges to typical reactions based on each of the three governance styles, we might have a better idea how to tackle these challenges. Table 1 links some of the problems observed with regard to the application of the EU SEA Directive (Meuleman, forthcoming) to reflexes which are typical from the perspective of different ideal-typical governance styles.

Table 1 Simplified heuristic model of SEA problems and their possible solutions by governance style (own composition).

SEA problem	Governance topics	Hierarchical gov. reflex	Network gov. reflex	Market gov. reflex
Insufficient scoping	Dealing with uncertainty	Scoping is the responsibility of government	Include stakeholders in scoping phase	Scoping is the responsibility of the planning agency
No cumulative and synergistic effects	Complexity	Keep it manageable and facts-based	Create consent on difficult topics like cumulation and synergy	Keep it simple, focus on direct impacts
	Dealing with uncertainty
Insufficient consideration of alternatives and scenarios	Reflexivity	Rule of law approach: focus on existing legal provisions	Maximalist approach: open to alternatives, including those proposed by stakeholders	Flexible attitude but ruled by cost of assessing alternatives
	Compliance
	Flexibility
Insufficient quality of the IA report	Reporting	Focus on statistics	Focus on consensus on data	Focus on monetisation
	Review
Insufficient explanation of uncertainties and other difficulties	Complexity	Rule out uncertainty by defining a narrow scope of the plan	Joint fact finding and participative scenario process	Select no regret options
	Dealing with uncertainty
Unclear impact of public participation	Participation	Public will be informed in due time	Public should be informed during the whole process	Public should be informed on a need-to-know basis
Insufficient consideration given to monitoring	Monitoring	Monitor impacts of measures	Monitor change	Monitor costs and benefits
Insufficient capacity of authorities	Capacity building	Establish a quality control authority	Establish a learning community	Establish a competitive training centre

Three underlying IA problem fields related to weaknesses and strengths of governance styles are elaborated below: different views (1) on the usability of knowledge, (2) on data collection and data use, and (3) on value-based conflicts.

3.2.2. Usable knowledge

The three ideal-typical governance styles imply different views on what is ‘usable knowledge’ for decision-making. For a decision prepared in a context of hierarchical governance, authoritative and undisputed knowledge is highly valued. In a context of market governance, costs are considered crucial, which may increase pressure to speed up the process and rely on relatively simple models. In a context of network governance (dialogue and appreciation of different views), consensus on knowledge is likely to be valued higher than scientific authority (Meuleman 2012). A comparison of five case studies of complex environmental and infrastructural problems in the Netherlands (In't Veld 2000) showed that a relatively rational, authoritative way of dealing with the knowledge which should support political decisions led to long delays and public anger. The authors argue that in such cases the IA should be more inclusive (joint fact-finding processes). Finally, knowledge disputes may emerge during IA processes because of the different paradigms of IA scholars and practitioners coming from different academic disciplines, such as law, economics, geography, engineering and natural and social sciences. Some may have, by training, a stronger preference for using statistics and utilise a different definition of what is ‘usable knowledge’ for political decisions than political scientists.

3.2.3. IA data collection and data use: governance, numbers and knowledge

The three ideal-typical governance styles have ‘incompatible contentions about what is knowable in the social world and what does or can exist’ (Dixon & Dogan 2002). This results in contrasting and mutually undermining convictions about the meaning of numbers in IAs: hierarchical governors consider numbers to be objective (a positivist attitude); market governors favour a ‘rational choice’ approach in which monetising is needed to get ‘objective’ information (Federickson & Smith 2003); network governors tend to emphasise the boundedness of rationality and have a greater appreciation for qualitative assessments.

A closer look shows that hierarchical governance is consistent with a preference for clear rules, formal procedures and clear problem definitions, while at the same time tendentially disregarding complexity. Faced with an IA for a plan or project that has fuzzy, ‘intangible’ or complex outcomes, hierarchical decision-makers may attempt to invoke urgency (‘no time for dialogue, we have to act now’) or fragment a project which results in neglecting cumulative effects (for example, ecosystem interactions or social subsistence), complexity, unpredictability and uncertainty. Such a rational approach may conflict with a more strategic approach to, for instance, SEA: ‘What makes SEA practice fuzzy is not the context per se but the use of technical-rational evaluation models in decision-making contexts that do not behave according to those rationales’ (Jiliberto 2012). The rational approach builds on the assumption that there is one single and central decision-maker and a clearly defined decision process, that the consequences of decisions can be reasonably predicted and that providing information about consequences of a decision is enough to make ‘better’ decisions (Monteiro & Partidário 2012).

A relatively strong appearance of network governance within the governance system – as observed in the Netherlands and Scandinavian countries for ages (e.g. Kickert 1997) – is globally rather the exception, at least at national level. One would normally expect here collaborative forms of data collection and interpretation (joint fact finding), but the Dutch example mentioned above (In't Veld 2000) shows that reality may beat theory.

The focus on numbers and monetisation in market governance may result in IAs which are relatively weak in qualitative analyses and which use cost–benefit analysis (CBA) rather than cost-effectiveness analysis (CEA) or multi-criteria analysis (MCA). IAs with a focus on monetisation rather than on qualitative and participatory approaches risk to neglect environmental, social and other non-market considerations (OECD 2010, p. 5). This may, among others, result in lack of public support for the implementation of the final decision.

It is a political reality that many IAs are carried out in a context of hierarchical and/or market governance. Such a context matches with strong confidence in statistics. An example of an underlying preference for quantitative data is that, although the EU comprehensive IA Guidelines explain that such data are not always available, the Guidelines state that ‘the more quantification you can provide, the more convincing the analysis will generally be’ (European Commission 2009, p. 32). As CBAs will always produce quantified information, it is no surprise that they have become fashionable within comprehensive IA methodology, regardless of the accuracy of the produced numbers (Niestroy 2008). The EU IA Guidelines advise that CBA, CEA and MCA, or combinations, may be used. CBA is often criticised because it does not cover impacts which cannot be monetised, as such impacts are abundant among environmental and social impacts. Some authors therefore consider CEA and MCA being more feasible for encompassing such impacts than CBA (Ackerman & Heinzerling 2004). The choice of instruments differs with the scope of IAs: CBA seems to play a more prominent role in comprehensive IAs than in environmental IAs (SEA, EIA), probably because economic impacts are not central in the latter. The increasing acceptance by decision-makers of behavioural sciences, including behavioural economics (see e.g. Thaler & Sunstein 2004; Kahnemann 2011), which show that what people do follows various precise logics but is only to a limited extent ‘objectively’ rational, might lead to a decrease of the importance of rational methods in IA. Nevertheless, Lobos and Partidário (2014) observe that the technical-rational model of environmental assessment, since the beginning typical for EIA, still dominates both EIA and SEA practice.

Dealing with uncertainty and unpredictability does not have to be unstructured. Funtowicz and Ravetz (1994) already presented the NUSAP model,10 which enables different sorts of uncertainty (such as uncertainty on causal effect, insufficient lack of data, lack of information on probability – which implies lack of predictability) in quantitative information to be displayed in a standardised and self-explanatory way, and thus provides clarity about its uncertainties. Uncertainty therefore requires being ‘precise about imprecision’ (Kaufmann et al. 2000): There is indeed no reason why qualitative assessments should not be carried out as rigorously as quantitative assessments.

3.2.4. IA and value-based governance conflicts

Governments seeking public support for their decisions are usually well advised to invest in the ‘trust’ value. Hierarchical governance is, however, a low-trust style. When this style dominates the governance of a policy initiative, decision-makers may, democratically legitimised, act in ways which increase distrust of stakeholders and the general public of the decision process and the decision-makers. Examples are IA procedures with inadequate publicity, restricted access to documents, short consultation deadlines, hearings organised as only information meetings, or which discredit evidence from stakeholders as ‘not authoritative’. Such procedures may have the advantage of saving time, but may also undermine public support or cause additional litigation, delays and costs afterwards, depending also on the government system: in non-democratic countries, public support may not be an important variable and litigation may not exist.

Different value systems – on many levels: national/regional/local/organisational/personal – give rise to different styles of governance. Combined with different views on knowledge disputes, this is a frequent cause of conflict. First, it has been argued that value-based conflicts do not require analytical but political problem-solving (Lindblom & Cohen 1979). This applies especially to the relations between multi-level and transboundary governance and IAs. The concept multi-level governance (MLG) concerns policy-making activity performed within and across politico-administrative institutions located at different territorial levels (Stephenson 2013). MLG has a hierarchical component (the formal relations between levels, power distribution, gaps of overlaps of competences, lack of coherence, different administrative realities) as well as a more informal one (voluntary collaboration between levels, different cultures and traditions). A well-functioning MLG system could be beneficial for IAs, as it would ensure adequate exchange of information between the levels. Kull (2014) investigated several EU policy cases, showing that reservations by the higher levels of government in such an MLG context about delegating functions and empowering subordinate levels depend on their administrative culture and the values and development visions of individuals in key positions. To address value conflicts in MLG, Jessop (2004) introduced a meta-level of analysis and design: multilevel metagovernance.

Value conflicts may hinder transboundary policies, plans and projects, which are covered by, for example, the UNECE Espoo Convention on EIA and its Protocol on SEA. The UNECE Aarhus Convention aims to ensure access to justice and information, but if there is no trust between parties on both sides of a border, transboundary IAs may be difficult to carry out. Both Conventions and the SEA Protocol supply basic governance arrangements which are necessary to implement SEA/EIAs: cross-border consultation, information and access to justice. A case in point is when countries have troubled or absent diplomatic relations; international IA agreements are then toothless tigers. In such cases, the challenge is to re-establish collaboration and communication without the ambition to (re)establishing diplomatic relationships between countries (Kolar-Planinsic et al. 2013).

A specific type of value-based tensions concerns relational values which determine how people relate to the values of other people (In't Veld 2013). Although the traditions and norms upon which beliefs are based are often taken for granted and hence not reflected upon, people tend to find their own values the best in the world. How people value other people's values varies from wanting to destroy them to fully accepting them as (also) useful. The point is that different governance styles are consistent with different relational values, and this makes it also relevant for IA practice. The relational value of an IA manager might be hegemonic (hierarchical governance), indifferent or autonomous (market governance) or tolerant or pluralist (network governance). This will have an impact on how the IA process is conducted, for example with respect to how consultation and participation are designed, and how public support for a government decision is achieved.

Risk assessment, a crucial part of IAs, is also affected by the primary values in the governance environment. Because, generally speaking, network governance is more open to complexity and unpredictability, Renn (2009) pleads for inclusive risk governance, as

the integration of knowledge and values can best be accomplished by involving those actors in the decision-making process that are able to contribute all the respective knowledge as well as the variability of values necessary to make effective, efficient, fair and morally acceptable decisions about risk.

His argument implies that risk assessments which are carried out in a governance environment which is predominantly rational (hierarchical and market governance) will not allow such an inclusive approach.

3.3. IA contributions to accommodate governance systems

The next step in the analysis is to explore how IA may contribute to the goals of each of the three ideal-typical governance styles. Rather than discussing the styles separately, I will briefly address two broad themes here: (1) matching problem definition and governance style and (2) matching evidence, transparency and participation.

IA can help matching the governance style with the type of societal problems which are believed to need addressing. It is known that hierarchical governance is an adequate style for crisis management, network governance for complex, ‘wicked’ problems, and market governance for routine problems (Meuleman 2008). However, societal problems are social constructs: there are many ways of framing the same problem. Therefore, IA practitioners can, on the one hand, try to increase the efficacy of IA for a specific governance environment by framing the assessed impacts in which is congruent with the terminology of the dominant governance style. On the other hand, they may wording use a more confrontational approach by commenting on the adequacy of the chosen problem definition, as each governance style has blind spots in terms of problem types. For example, proponents of hierarchical governance, who believe that command and control are important, may tend to see critical/urgent problems everywhere and ‘wicked’ problems nowhere. Proponents of network governance, who find consensus and extended reflection important, may see every problem as a wicked problem, but at the same time may ignore crises and avoid routine solutions. Proponents of market governance tend to advocate rational relations between cause and effect, as well as standard solutions, which may lead to wrong choices during crisis situations and denial of wicked problems.

If the governance context promotes this, IA can lead to better evidence, transparency and participation in the policy process. Jiliberto (2012) argues that SEA indeed helps to improve the legitimacy of strategic decisions and broadens the range of actors participating in them, that it promotes a strategic view in recognising society's environmental values and that it helps to improve the quality and accountability of their decisions, while respecting the existing legal framework and equality for all parties in disputes. These are indeed elements of the ‘New Public Management’ governance mixture which dominated in many (Western) countries around 2000, around which time the European SEA Directive was established.

As regards participation, SEA clearly seems to support collaborative governance processes, at least when SEA practitioners ‘translate its principles in accordance to the general principles of successful collaborative governance and joint fact finding’ (Van Buuren & Nooteboom 2010). IFIs such as the European Investment Bank, the European Bank for Reconstruction and Development, and the World Bank have established the practice to require that countries as part of a lending agreement carry out SEA, EIAs or other IAs. The results include that in some developing countries IAs have become good practice even if there is no national legal IA system in place. While such IAs are being carried out, governance principles like public consultation, transparency, administrative capacity, coherence and effectiveness are stimulated and may become more widely applied as part of a nation's governance system.

4. Improving IA governance within the wider governance environment

Because the governance environment influences IA practice, IAG should take into account the constraints and opportunities of the wider governance environment. IAG can be defined as setting up and managing the application of formalised IA instruments in relation to actors, interests, legal system, and culture and tradition. It includes deciding who has responsibility for what, who to involve and how, what are relevant linkages and collaborative approaches, and which assessment methods and models (including their underlying assumptions) will be chosen. IA governors are those who set up, manage or implement IA systems: this includes all IA practitioners in administrations, companies or consultancies. Based on the analysis in this article so far, four action-oriented ‘principles’ for improving IAG are suggested below.

4.1. Principle 1: organise reflexivity

IAG takes place within social systems. Social systems are reflexive (e.g. Giddens 1991): they constantly adapt to new circumstances, including to assessments or predictions of impacts in IAs. Legal and administrative systems, in contrast, tend to resist to external influences. Together with the impediment that law-making often takes many years, the result is that legislation is based on theories of the past, and often, by the time a law enters into force, the problem is solved or has changed its shape. For IA practice it is relevant that every policy or legislative proposal is based on an (often implicit) ‘policy theory’, a set of implicit social and behavioural assumptions that underlies a public policy (Leeuw 1991). Such a theory expresses expected causalities between means, instruments and objectives. For example, when the dominant policy theory considers a problem as traffic congestion, the preferred solution may be more roads. When the theory says that the problem is that inner cities cannot be reached, the solution may be largely based on public transport. When the solution to a problem is a law or plan which enables new major road infrastructure, the making and implementation of such a law or plan often takes so many years that meanwhile there is another policy theory. Policy theories may align with governance styles: a hierarchical style will rather embrace legal than voluntary solutions to a problem. Not all project initiators or competent authorities are conscious of their own governance style and policy theories and the possible incompatibility with the challenge that is discussed during the IA process. This may be reflected in the unwillingness to investigate alternatives in the IA.

4.2. Principle 2: analyse the governance environment

In section 2 it was argued that the governance environment of any particular IA includes the whole of rules and laws, institutional setting, policy instruments, division of tasks and roles of governments and societal stakeholders, within a particular country or other boundary. It is important to ensure that IAG (e.g. choice of instruments, level of participation and quality of knowledge) is compatible with the governance context of the plan/project, in order to prevent or mitigate conflicts. There are specific tools for analysing the aspects of the governance environment as introduced in Section 2: which actors are relevant (including governmental and non-governmental actors), which standpoints and interests they have, how they argue (argumentation/discourse analysis) or how they relate to each other (strength analysis, relation/network analysis) (for a description of these tools, see e.g. Meuleman 2003). Part of the analysis of the governance environment is to determine the (main) governance style with which the project will be executed, and to try to match the knowledge produced in the IA process to the type of knowledge that will be important to decision-makers.

4.3. Principle 3: complement and switch between governance styles

A typical metagovernance approach to IA would be to include tools from other styles if one governance style dominates in the design of the IA process, as this can bring about a more rounded approach. For example, dialogues (network governance) need some sort of structure (hierarchical governance), efficiency (market governance) is useless without effectiveness (hierarchy), and authority (hierarchy) erodes without trust (network). Arts and Faith-Ell (2012) propose, for example, to introduce elements of market governance such as public–private partnerships, green public procurement and contracting in infrastructure governance, to strengthen the relatively weak position of the legal instrument of EIA (in this sense an instrument of hierarchical governance) to make a project more sustainable. Other examples of market governance instruments complementing hierarchical governance include codes of conduct, eco-labels, certification schemes and voluntary reporting channels. Empirical examples have also shown that metagovernance through combining network and hierarchical approaches can create a breakthrough in (environmental) conflicts (Parkins 2008). Another metagovernance strategy is switching to another governance style when circumstances change. Sometimes the phases of a large development process trigger a ‘natural’ switch to a different governance style (see e.g. Lowndes & Skelcher 1998, on urban renewal projects), and when IA is meant to accompany all phases, IAG should be reflexive enough to adapt, e.g., the type of participation to new phases.

4.4. Principle 4: organise appropriate public consultation and participation

Public consultation and participation can enrich IA processes with better data, insights and more support, under the condition that it is organised in an appropriate way, appropriate within the context of the policy, plan or project and its governance environment. This does not always imply maximum participation: sometimes less is more. Appropriate consultation and participation should include staying within the limits of acceptability of the governance system – unless, of course, the IA is used as governance intervention to stretch the existing limits. It should also be timely. One important variable, in many cases supported by legal obligations, is whether to consult the public in an early stage of a policy/plan/project. Some argue that too early consultation is meaningless, because the plans will be not yet concrete enough. One way to get more out of such an early phase is to turn consultation into participation or even co-production: a ‘mutual gains approach’ (MGA) may help increase support for and ownership of the initiative. The Netherlands' Environment Ministry used to be a keen supporter of this approach and during the 1990s required all its policy officers to attend a three-day MGA training. A classical problem with all consultation and participation that is not specific for IA is that it is sometimes difficult to find a representative ‘public’. Among people who voluntarily participate in public consultation processes, privileged actors may have a strong voice, stronger than they would have had based on democratic representation. Social media and the Internet have made the emergence possible of sometimes influential NGOs which in reality represent only one person or one company. In countries where speaking up is culturally risky or simply not done, the imbalance between a dominant ‘vocal minority’ and the ‘silent majority’ might be even stronger than elsewhere (Peirson-Smith 2013).

5. Conclusions

This article reviewed the relationship between concepts and practice of IA and governance. On the questions raised it can be concluded:

• It makes sense to think seriously about governance when IA is carried out, as governance systems offer both constraints and opportunities for the governance of IA systems and procedures. Improving IAG requires sufficient understanding of the governance environment, including its implicit values and traditions.

• Linking IA and governance sheds new light on a range of existing problems with the design and implementation of IA systems. First, because IA problems can be related to typical weaknesses of governance styles. Examples include conflicts around data collection and data use, and value-based conflicts. Second, because governance systems may offer opportunities to strengthening IA systems through, e.g., economic recovery governance. The example of the European Semester illustrated this. Acknowledgement of the relational and normative dimensions of governance and of the dynamics of combining different governance styles (metagovernance) enables a broad perspective on the challenges of IAG.

• Taking the opposite perspective, looking from IA at governance, it was shown that IA can contribute to the discourse on what the most effective framing of problems (and possible solutions) is. It can also lead to better evidence, transparency and public consultation, when the governance environment is open to this. Finally, introducing IA in a country might result in a more inclusive governance environment.

• Understanding the governance environment and using principles from governance theory such as reflexivity and contextuality, and metagovernance strategies (complementing governance styles and switching between styles), can help improving the governance of IA.

To conclude, it looks as if IAG develops into a new area for transdisciplinary research and applied studies, with fascinating perspectives for both IA and governance theory and practice. There are already interesting research projects (e.g. Monteiro et al. 2014) and studies (e.g. on the potential of IA in the context of the EU economic governance) under way. IA and governance are on the same ground. The owl's wisdom and the beehive's organisational talent need each other.

Acknowledgements

The author would like to thank the IAPA peer reviewers and the following experts for their valuable comments: Stamatios Christopoulos, Michael Kull, Elena Montani, Margarida Monteiro, Ingeborg Niestroy, Jonathan Parker and Maria Partidário.

Notes

This article elaborates on the two-page Fastips No 4 on Governance (Meuleman 2013b) of IAIA, which was a first attempt to formulate essential relations between IA and governance. The article does not represent the opinion of the European Commission.

^1. ‘Intelligence’ is used here with the American English double meaning of ‘information’ and ‘brainpower’.

^2. Quote from http://wbi.worldbank.org/wbi/topic/governance, retrieved on 16 June 2014.

^3. Sustainability IA assesses economic, social and environmental impacts and their relations.

^4. With comprehensive IA, I mean here the EU IA system which covers regulatory, environmental, economic and social impacts.

^5.http://ec.europa.eu/europe2020/making-it-happen/index_en.htm

^6. The Annual Growth Strategy, National Reform Programmes and Country Specific Recommendations, supported by Commission Staff Working Documents.

^7. Article 2.3 of the 7th EAP. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri = OJ:L:2013:354:0171:0200:EN:PDF

^8. European Environment Agency (2013).

^9. Political scientists tend to use this (historically disputable) term for the five liberal economies where market governance (in the form of ‘New Public Management’) originated and in which this style became dominant: Australia, New Zealand, Canada, UK and USA (see e.g. Hood 1991).

^10. NUSAP is an acronym for the five categories ‘Numeral’, ‘Unit’, ‘Spread’, ‘Assessment’ and ‘Pedigree’.