I. Introduction

Consultation is part of a broader set of tools to ensure stakeholder participation in the activities of both public and private institutions. Various forms of engagement may reflect different perspectives of citizens’ participation. In relation to public institutions, stakeholders’ engagement should be granted at all of State, regional, and international levels.Footnote ² Consultations broaden the active involvement of citizens in the exercise of institutional powers: legislation, administration, adjudication. As to private institutions engaged in rule-making participation, the consultation process can ensure legitimacy and improve effectiveness.

Stakeholders’ engagement should be pursued not only at the level of policy design but also of policy implementation and evaluation, as it pertains to the entire policy cycle.Footnote ³ Growing concerns focus on the accessibility and exclusion of large segments of civil society from effective participation in both the legislative and the administrative process.Footnote ⁴ It is still a matter of debate what constitutes inclusive policy-making and which tools should be used to ensure effective participation by those ultimately affected by the decisions.Footnote ⁵ However, more attention should be devoted to the different workings of the various tools that can be deployed in the consultation processes.

Principles concerning open government, defining citizens’ access to legislative and administrative procedures, have been put forth in many charters and declarations. There is general consensus over the core principles, which should include participation, inclusiveness, openness, effectiveness, and coherence.Footnote ⁶ However, divergences emerge on how these principles should be applied. The principles are not necessarily always consistent with one another, hence trade-off decisions may be needed, for example between inclusion, effectiveness, and coherence. These principles, and the trade-off decisions occurring in their implementation, greatly affect consultation design. We propose a conceptual framework for consultation design that incorporates incentives for participation (effectiveness) and adequate consideration of cognitive (in)abilities of potential participants (inclusiveness).

To examine the issue we deploy behavioural sciences, both in the perspective of behavioural game theory (to study intrinsic motivation to participate) and in that of behavioural science (to study the impact of heuristics and bias on the consultation procedure). The procedural dimension of consultation has not been fully investigated in light of behavioural sciences. The democratic values underlying open government may be undermined by an architecture of participation that does not warrant inclusion of groups and communities with lower resources and cognitive abilities.Footnote ⁷ Hence, here the definition of stakeholder engagement is more on implementation than on principles. Mapping stakeholders ex ante becomes a strategically significant element of the consultation architecture in order to ensure that relevant interests are duly taken into consideration. Stakeholder mapping therefore goes beyond being a mere descriptive exercise, as it possesses an important normative character.Footnote ⁸ While in this article we concentrate on the issue of stakeholder mapping and its behavioral relevance for the consultation process, we are aware that a behavioural approach to consultation is broader than what we put forth here. Next to the strategic and cognitive behavioural components that we analyse in this paper there are at least two more fundamental behavioural sides of consultation design that we plan to cover in future research. First, the choice of tools in specific consultation design should be motivated by behavioural considerations. For instance, a specific consultation tool may be thought of as more efficacious to counter cognitive biases that are expected to appear among stakeholders, therefore it should be tested to discover whether it would in fact be efficacious, and hence deployed if the test were positive. Second, the issue of cognitive bias needs to be taken into account not only in connection with shareholders, but also in connection with the administration, in that administrative decision-making processes should also be analysed through a behavioural lens, as its agents are victims of cognitive bias just as much as stakeholders are.Footnote ⁹

Stakeholder engagement is not only an issue for EU and international public institutions but it also characterises the evolution of transnational and domestic private regulation.Footnote ¹⁰ Global private standard setters have made significant progress in defining procedures that involve stakeholders in standard setting, monitoring, and enforcement.Footnote ¹¹ Differences emerge between single and multistakeholder organisations where the framing of the consultation may reflect the co-existence of divergent views about the new policy.Footnote ¹² A comparative analysis of consultation in public and private standard setting processes is beyond the scope of this paper. It is, however, worth noticing that there has been mutual influence between public and private organisations when promoting stakeholders’ engagement, but differences in instruments and effectiveness are significant. In fact, the engagement process can derail and lead to outcomes of regulatory capture. It is therefore paramount that both at the level of economic incentives and at the level of intrinsic motivation, the architecture of consultation mechanisms take into account capture risks and deploy tools that presumably would reduce or eliminate such risks. Instead of focusing on material incentives to rule out capture, in this paper we focus on intrinsic incentives and explore their interplay with cognitive factors; in so doing, we purport to introduce new tools to better protect weaker stakeholders. For instance, considerations related to the cognitive abilities of stakeholders might reveal that some of them are in a weaker position than they were otherwise assumed to be, and therefore require better engagement promotion. Or, consider that engagement relies particularly on intrinsic motivation, especially when it comes to stakeholders that are affected by policy decisions and yet lack a significant voice if not participating en masse in the process. Thus, by focusing on intrinsic motivation and cognitive abilities, we highlight both new possible risks of regulatory captures and new possible remedies.

This article is structured as follows: After this introductory section, in section II we provide a principled justification for the adoption of a dialogical architecture in the consultation procedure. Section III details the behavioural point of view of our analysis, covering (section III.1) the heuristic and bias themes relevant to consultation and (section III.2) the behavioural game-theoretic aspects in both dimensions of the dialogue between administration and stakeholders and of the dialogue among stakeholders. The strategic structure is formally described in section IV, while section V puts forth three dialogical models summing up the elements elaborated in the previous sections. Section VI counters a possible objection and section VII concludes with policy recommendations.

II. A dialogical approach to consultation

There is no conceptual consensus over the distinction between consultation and other forms of stakeholder engagement.Footnote ¹³ Consultation differs from information transfer and from other forms of stakeholder engagement.Footnote ¹⁴ It is generally defined as an instrument to engage stakeholders and ensure their active participation.Footnote ¹⁵

Consultation design may affect the ability to participate in the decision-making process and influence the final decision made by the consulting body. Publishing the document on a website is a necessary but not sufficient condition to ensure inclusive and effective participation. Both in open-to-the-public and in targeted consultations, the procedural features may affect the effective participation. For instance, in targeted consultations, where only a selected group of stakeholders are involved, using several consultation instruments for different groups of stakeholders, based on the specific cognitive resources of stakeholders, may promote participation and result in an overall better quality outcome. Consultation may aim at acquiring new information and/or at testing predefined hypotheses. In the former case information can come from experts or stakeholders. The process of information acquisition includes technical information originating from experts and non-technical information originating from interest groups like industries and civil society organisations. Our analysis focuses on stakeholders and will not address consultation design when experts are the target.

The definition of objectives, the stakeholder mapping, the correlation between various tools and classes of stakeholders, the duration and the number of rounds, all are strategic characteristics of a consultation procedure.Footnote ¹⁶ When mapping stakeholders, a key distinction is one between those who will implement the regulation and those who are affected by the results of the regulation. The position of the administration may differ in comparison to that of the various participants. In relation to some stakeholders the administration can be less informed (often this is the case with industry) and needs to acquire information. In relation to other stakeholders it can be more informed and the consultation process should aim at generating their cognitive empowerment by transferring information to the less informed.Footnote ¹⁷ These observations suggest that two goals be incorporated in the design of the consultation process. In the former case, the consultation design should aim at reducing the information asymmetry so that the administration can make a better informed decision. In the latter case, the design should aim at enabling the (potentially) negatively-affected and less-informed stakeholders to learn and overcome their cognitive limitations. These different objectives may influence the procedural design and the dialogical structure of the consultation. They may also call for a different stakeholder mapping for policy design and policy evaluation. Hence both consultation design and the identification of stakeholders may change from the ex ante to the ex post stage.

Mapping is crucial for identifying groups that might have interest and power to impact the policy initiative and groups that may only be affected by the policy at stake without being in a position to fully influence it. But even more important is to make sure that groups representing conflicting interests participate in the consultation in proportion to their “social weight”,Footnote ¹⁸ in order to guarantee that divergent views can all be taken into account by the administration. A fair chance to participate means that each relevant stakeholder group should be given an opportunity to make their opinion heard. However, stakeholders’ relative weight may vary according to their influence and the disproportionate participation of one constituency compared to others may have a stronger impact, unduly influencing the outcome of the consultation process.Footnote ¹⁹

The nature of the stakeholders’ interests can affect the cooperative or competitive structure of the consultation process and of its design. Stakeholders’ interests may be aligned or conflicting. Hence grouping stakeholders according to the degree of interests’ homogeneity may have procedural implications for consultation design. A dialogue among conflicting stakeholders may have different outcomes from a dialogue between homogeneous stakeholders.

What should then drive the choice of instruments for stakeholders engagement? The choice of instruments should be correlated to the various objectives of the consultation process and the degree of interests’ heterogeneity. The maximisation of stakeholders’ dialogue suggests the use of a single instrument where the different entities, individuals and organisations, can take part and engage. But heterogeneity of stakeholders based on different economic and cognitive endowments may lead to ineffective participation. Relevant variations in stakeholders’ endowments may require the use of different tools according to “homogenous” classes of stakeholders. We recommend to differentiate among stakeholders within the same consultation instrument to the extent possible. Only if the level of heterogeneity is too high do we suggest the adoption of different consultation tools. In the latter case they should be coordinated so that consistent findings can be drawn.

III. Designing the dialogues

The following analysis focuses on a particular type of consultation where the consulting administration decides: (1) the modes of participation in the consultation by differetnt classes of stakeholders; and (2) the degree of interaction between the participants. We do not examine fora or other interactive instruments where the administration acts more as a facilitator.Footnote ²⁰ We include in the analysis both open access and targeted consultation.Footnote ²¹ We shall address the issue of what kind of interaction takes place in a consultation, argue that it is a strategic interaction of the kind studied by the theory of games and frame our analysis in game-theoretic terms.

Consultation, we have said, may pursue different objectives: gaining legitimacy, acquiring information, promoting stakeholder dialogue. In this analysis we focus on the production and transfer of information from stakeholders to the administration and the exchange of information among stakeholders. We consider two potentially combined dialogues: between administration and participants, and among participants.

The creation of a dialogue between the administration and the participants – through the design of the consultation procedure – ultimately pursues a twofold, intertwined, objective: (1) producing and transferring information; and (2) influencing the final decision of the administration. The dialogue between stakeholders aims at ensuring a level playing field to influence the administration’s final decision (as aim (2) invokes issues of equality of access) and at redistributing knowledge among participants (which, in turn, can foster aim (1) above). The dialogue can also affect participants’ views and change their positions during the consultation procedure.

Participants in the consultation can be thought of as a “community” whose identity is ex ante unknown by the administration since the decision to take part in the consultation is always voluntary and depends on the incentive structure. However, this community is often composed by repeat players who take part in various consultations over time. The degree of knowledge between participants may determine whether they should be considered as partners or strangers and partly define the degree of mutual trust and the incentives to produce new information.Footnote ²² The “community” is also composed of members whose interests are usually conflicting and the consultation design must ensure equal opportunity to access and express their views. A third dimension of community’s governance in consultation procedures is the relationship between individual and collective interests.

The aim to generate new information can be framed within two dialogical structures that need to be kept separate though coordinated: One is the dialogue between the party designing and implementing the consultation (hereinafter the administration) and the participants (the communityFootnote ²³ ). The other is the dialogue amongst the participants during and after the consultation. As we shall show, lumping them together or focusing only on the former dimension can generate procedural fallacies in consultation design.Footnote ²⁴ Such procedural fallacies can, in turn, undermine the objectives of consultation, generating biased information about stakeholders’ opinions.

IV. Cooperative and competitive interactions

In this section we detail the behavioural game-theoretic model on which the dialogical designs described in section III.2 above are based. The underlying assumption of our modelling strategy is that information becomes a public good through consultation, in that it influences the decision-making process of the administration. We assume that stakeholders’ opinions are trustworthy and that the decision maker makes the best use of the information it receives. We furthermore assume that the administration has initially an agnostic stance and it is linearly influenced by the information sent by the stakeholders (the players). Players’ payoffs are yielded by the gains made by saving search and production costs about information and by the benefits of shared information. The model varies along two dimensions: whether the interests of the stakeholders are convergent or divergent, and whether stakeholders express their opinion simultaneously (that is, ignoring other stakeholders’ opinion) or sequentially. The latter dimension is modelled straighforwardly through sequential and simultaneous games. We propose here to analyse the former through the use of two different kinds of public good game.

In the case of convergent interests, we have a typical public good game, in which stakeholders prefer that more, rather than less, information be released, but at the same time they have an individual incentive to free ride on the information supplied by others and keep their resources otherwise allocated. Players’ endowments represent, in general, their available resources. They may allocate their resources to the task of gathering, elaborating, and presenting information. The larger the amount of resources a stakeholder allocates to such tasks, the better the information received by the decision maker will be, but also the less resources will remain available to the stakeholder for other activities. Thus, providing good information comes at a cost (effort) and yields a public good in that it allows the decision maker to make an informed decision that presumably will benefit all stakeholders. More precisely, players have a certain initial endowment that they can fully or partially invest into a public good. Once all players have decided how much to contribute to the public good, the sum total of the contributions is multiplied by a constant term and divided among all players. This reflects the fact that the information produces benefits for all participants. More formally, let us say that there are n stakeholders (players), each one with a symmetrical endowment of e units. Each player i=1,..., n offers a contribution c _i to the public good. The sum of all contributions is then multiplied by a constant factor f < n, and divided by n players. Thus, the payoff function for player i is

$$u_{i} (c_{i} ){\equals}e{\minus}c_{i} {\plus}k\mathop \sum\limits_{i{\equals}1}^n c_{i} $$

with 0 ≤ k ≤ 1.

In the (basic) case of divergent interests, we have two groups of stakeholders each trying to build their own public good, supplying information that pushes their own agenda with the decision maker. In this case the free riding problem lies within each group of stakeholders. The model here is a competitive public good game,Footnote ⁴⁴ in which two or more groups of stakeholders compete to influence the decison maker according to their diverget interests. Endowment and effort are understood as before, but the composition of the public good reflects the divergence of interests at stake. Indeed, the public good in this case is the result of the competition among groups of stakeholders, mimicking the intuition that the stakeholders exert more influence on the decision maker with the more effort they spend to participate in the consultation process. In this case, each group works for the creation of their own public good. The competitive element lies in the fact that however large the amount of public good one group has created, it can be destroyed by the influence on the decison maker exerted by the opposing group, since the effort of the other group influences the administration in the opposite way. This is captured by producing the public good as the difference between the sum total of the contributions of each group. Formally, in this case we have two groups A and B of m stakeholders each, each stakeholder with the same endowment of e units. Each player i _X (with i = 1,..., m, X = A, B) offers a contribution $c_{{i_{X} }} $ to the public good pertaining to group X. The overall public good, which is multiplied by a constant factor f < 2m and divided by 2m players, consists of the difference between the total contributions of one group and the total contributions of the other group. This, the payoff function for player i _X is

$$u_{{{A}_{i} }} (c_{{{A}_{i} }} ){\equals}e{\minus}c_{{{A}_{i} }} {\plus}k\left( {\mathop \sum\limits_{i{\equals}1}^n c_{{{A}_{i} }} {\minus}\mathop \sum\limits_{i{\equals}1}^n c_{{{B}_{i} }} } \right)$$

with 0 ≤ k ≤ 1.

Given this setup, what is the strategic analysis of a competitive public good game (CPGG)? We can think of a CPGG as a collection of social dilemmas played within one group, one for each contribution level of the other group. The overall payoffs of the collection of social dilemmas will be higher when the sum total contribution of the other group is lower. Thus, if the group A competing against the group B slacks and does not put in much effort, then, other things being equal, the payoffs in group B will be higher than if group A had put in more effort. The strategic element in a CPGG, therefore, remains the same as in a cooperative public good game: contributing 0 is the dominant action, as the return on positive contributions is fractional. Also, the entire group contributing the highest possible amount is Pareto-efficient regardless of the contribution level of the other group. And if all players contribute 0, they are better off than if all players contribute 10. Thus, while players still have an incentive to free ride, divergent and convergent interests define different objectives of the groups participating in the consultation. Both the content and the effort to produce information are affected by the divergent or convergent nature of the participants’ interests. Moreover, there is a psychological difference between the two incentive structures. We surmise that competitiveness may hinder tendencies to free ride internally and create intrinsic motivation to contribute, bringing about a more significant stream of information in the consultation process.

Whether the interaction yields a larger provision of information when it is cooperative (first case, convergent interests) or competitive (second case, diverging interests) is of course an empirical question, whose answer has important ramifications when it comes to identifying the best consultation mechanism to be implemented. Depending on stakeholder mapping, one can imagine the administration designing consultations with convergent or divergent interests with the goal of increasing or decreasing cooperation in the resulting interaction. The goal would depend on stakeholder mapping in the sense that stakeholders deserving a higher degree of protection (for instance those who are heavily affected by the policy at stake, yet have little power to influence the decision) should be part of a consultation process that increases their chance of being heard.

There is then the second crucial aspect in the context of consultation mechanism design, that is the question as to whether revealing information simultaneously at the end of the process (having participating stakeholders ignore the information revealed by other stakeholders) increases information provision, or whether the mechanism in which information is revealed “as it comes” so that stakeholders are, or may be, fully informed about the contributions of those who participated before fosters information provision more efficiently. It could be that the answer to the latter question depends on the former, in that when interests converge, simultaneous proceedings provide more information, while when interests diverge, sequential proceedings should be preferred. As we stated, these are empirical questions deserving empirical investigation, since their answers are highly relevant to the issue of the design of consultation procedures.

Summing up: from the theoretical point of view, in all four permutations yielded by the two dimensions we are considering, contributing anything to the public good is a strictly dominated strategyFootnote ⁴⁵ and hence never played by rational agents.Footnote ⁴⁶ We have reason to believe, however, that sequential and competitive setups yield a larger transfer of information to the administration than simultaneous and cooperative ones. Empirically, for instance, we know that in public good gamesFootnote ⁴⁷ rather high levels of cooperation are observed at the onset, while levels decay during play, suggesting that in an open consultation without publication of the opinions, participants’ cooperation could falter. Such decay is mitigated in sequential situations, where a kind of dialogue is introduced. Moreover there is some evidenceFootnote ⁴⁸ that competitive public good games trigger higher levels of cooperation than non-competitive ones, even though such evidence is restricted to the simultaneous case. Of course, there is a risk of crowding out cooperators, for instance in a sequential case in which cooperation is weak at the onset. Hence, it is paramount that the administration ensure the highest possible initial participation, in order to promote cooperation in subsequent rounds.

Subject to further research is the question concerning the potential divergence between the outcomes of the dialogue between administration and participants and that among participants. It is not necessarily the case that increasing the transfer of information to the administration corresponds to a more intense exchange among participants. Maximising at the same time the objectives of information transfer to the administration and exchange among participants might not be a viable strategy.

V. The models: procedural framing and the impact on dialogue effectiveness

We first address the time and sequence of publication of individual contributions to the consultation as one relevant design’s variable and offer three possible alternatives:

1. (1) the consultation procedure does not allow publication of the contributions during the consultation process but only at the end;

2. (2) the consultation process permits (if participants so desireFootnote ⁴⁹ ) to publish the contribution as soon they reach the administration. Potential participants can access the contributions and position themselves “against” the other participants during consultation. However they only have one opportunity, eg they cannot reply to the others in subsequent rounds;

3. (3) the consultation takes place in multiple rounds. The design determines the sequence of publications among different classes of stakeholders. In multiple rounds, parties may refer to what others have stated and engage them, increasing the dialogical structure of the interaction among participants.

The choice of the consultation structure, instruments, questions, the wording of the questionnaire are all elements that (cf section III.1) impact outcome and turnout in ways that behavioural sciences can usefully address. In particular, the models offer three different consultation procedures that, in our view, correspond to three distinct procedural frames. Framing is known, among other effects, to affect risk-propensities,Footnote ⁵⁰ to affect attitudes towards time-discountingFootnote ⁵¹ and attitudes towards future generations,Footnote ⁵² to cause preference reversals,Footnote ⁵³ to impact preferences more generally,Footnote ⁵⁴ or to induce more accepting behaviour in negotiations.Footnote ⁵⁵ Framing effects can be characterised in three classes:Footnote ⁵⁶

1. – risky choice framing (where an uncertain effect is described in terms of gains or losses);

2. – attribute framing (where a given attribute is described in positive or negative terms); and

3. – goal framing (in which the goal of a given decision is framed positively or negatively).

We surmise that in the case of consultations yet another kind of framing is relevant: procedural framing. Consultation results may vary depending on whether the consultation procedure is framed as a one-way, two-way, or multi-round dialogue. Although the behavioural impact of procedural framing has not been studied in the context of consultation, our game-theoretic underpinning of consultation procedures suggests possible behavioural effects relative to procedural framing. If, as we argued, crucial aspects of the consultation mechanism are to be understood as a public good game, the dialogical framing of the game affects quality and quantity of the information provided by the players (see section III.2). Furthermore, we need to consider the possibility that a particular dialogical frame may have a hindering effect on individual choice bias.Footnote ⁵⁷

VI. Different instruments for different stakeholders? How excessive “tailoring” may undermine the dialogical architecture

So far we have assumed that the consultation procedure does not differentiate consultation tools among stakeholders. The three models assume that only one instrument is to be used across different classes of stakeholders. However if stakeholders’ endowments are significantly different in terms of economic resources, competence, and cognitive capabilities it might be worth differentiating the consultation instruments.

Stakeholders mapping is generally made following three criteria: identify those potentially affected by the new instrument, those in charge to implement it, and those interested in the new policy.Footnote ⁶⁹ Neither economic nor cognitive endowments are taken into account according to the current schemes. We suggest incorporating endowments into the consultation design and correlate the choice of instruments with the endowments and/or the level of efforts parties are willing to spend.Footnote ⁷⁰

Usually, after mapping the relevant stakeholders, the administration in charge has to decide which instruments should be used for which categories of stakeholders. There might be a trade-off between the objective to promote stakeholders’ dialogue and the necessity to adapt instruments to the different stakeholders’ categories. A single instrument maximises stakeholders’ dialogue. Multiple separate instruments, tailored to the economic and cognitive abilities, can improve the quality of the participation but may reduce the effectiveness of consultation as an instrument to create a dialogue. The consultation designer has to choose on a case-by-case basis the degree of instruments’ differentiation that maximises the combined objective of building a dialogical architecture and increases individual abilities to contribute.

Once the map is defined then, a decision on the consultation tools should be taken. The consultation tool(s) depend upon the consultation method. The decision may fall on a single instrument or combine different ones according to the type of stakeholders potentially involved in the process.Footnote ⁷¹ For example, questionnaires may be combined with hearings when the involved stakeholders have very different incentives to provide information.

To the extent that a single instrument can take a modular form, it should be preferred to the use of multiple instruments directed at different classes of stakeholders. But modularity should preserve a common core of qustions. If such a common core cannot be kept, then it would be more effective to deploy various tools and coordinate them into a single consultation procedure. Clearly stakeholder dialogue in a framework of different instruments can be more difficult to operationalise. The role of administration as a vehicle of different approaches grows when multiple tools are in place.

VII. Conclusions and policy recommendations

Consultation design plays a significant role in ensuring that the objectives of stakeholders’ participation in legislative and administrative activities are achieved. We have identified two dialogical interactions: one between administration and stakeholders, and one among the participants in the consultation process. We suggest that behavioural biases and heuristics affect the responses, and that procedural framing should be accurately designed in order to ensure effective participation. We have examined two variables affecting the design: number of rounds and intensity of dialogue among participants. By no means are these the only relevant variables. Many others may influence the design and characterise the dialogical nature of the consultation. By focusing on those two variables we suggest using different models depending on whether the main objective is to acquire information or to promote stakeholders’ dialogue. Overall, we believe that one-round consultation with publications of contributions as they come can solve asymmetry problems with the administration. This model in the two variables 1 and 2, can be cost-effective when the main objective is to establish a dialogue between administration and stakeholders. When stakeholders’ dialogue is one of the main objectives of the consultation process, it is advisable to use multiple rounds (model 3) to target stakeholders, and to differentiate the questions according to cognitive capabilities and financial resources of contributors. The selection of the stakeholders’ group can be left to the contributors and having them choose the category to which they belong. For example, if there are different degrees of complexity, parties can freely select the one they belong to and use the instrument that is correlated to that class of stakeholders.

In multiple rounds, participation should be defined so as to maximise the quality and the efficacy of contributions by designing a dialogical architecture that can develop over time. The distinction between competitive or cooperative games should affect the design and the model of interaction among participants. When information provision is competitive in order to persuade the information, the level of effort will not be undermined by free riding problems. On the contrary, when information provision is cooperative it is likely that free riding and underprovision will be a problem that consultation design has to tackle.

The choice between models should be left to the administration, but more awareness about behavioural implications of consulation design should inform the consultation architecture. We strongly suggest that both European and national institutions issue specific guidelines concerning consultation design, taking duly into account behavioural variables of the consultation.