Towards a decision-support procedure to foster stakeholder involvement and acceptability of urban freight transport policies

Urban freight transport (UFT) is an important and under researched/undervalued topic policy-makers have to deal with. UFT is a complex world characterized by scarce knowledge and heterogeneous stakeholders with conflicting objectives [1–4]. Local policy-makers try to strike a balance between public and private interests, aiming at city sustainability while fostering freight distribution efficiency [5, 6]. Stakeholder engagement and ex-ante policy evaluations are two fundamental issues to be addressed to guarantee an effective and efficient decision-making process [7]. Stakeholder engagement implies that all the parties interested in the decisions are directly involved in the policy-making process. It is convincingly argued that better decisions are implemented with less conflict and more success when they are stakeholder-driven [8]. Taylor [1] identifies four fundamental stakeholders to be consulted when evaluating UFT policies: (1) shippers; (2) freight carriers; (3) residents; (4) planners/regulators. Shippers generate freight demand with the intent of either making final goods available to end consumers or alternatively to import/export semi-finished goods from/to other industrial partners. Planners/regulators define the overall framework under which transport providers perform delivery tasks. Public successfully implemented regulations are only possible when they are practically feasible for private stakeholders and help them pursuing their distinctive objectives. A dilemma often arises among the conflicting objectives of urban freight operators and their customers, on one side, and the community, on the other [9]. Private actors directly involved in urban freight distribution are mainly concerned with constraints, such as congestion or street/area-wide access restrictions, since these restrictions impose additional costs and delays on operations. Public actors involved in the oversight of urban freight distribution, as well as urban residents are, in general, concerned with congestion, air pollution, as well as overall city liveability [10]. There is usually a status quo bias in urban freight [11]: on the one side, city governments expect business to set up new logistic services fit to the emerging customers’ and retailers’ needs while also beneficial to the environment; on the other side, logisticians are expecting municipalities to initiate (and subsidize) new services before starting businesses which, otherwise, could prove poorly profitable and highly risky. Besides, Dablanc et al. [12] underline the importance of the territorial scale for stakeholder involvement to obtain an effective behaviour change: consultations are of little use at the local and municipal levels, if not combined with metropolitan or region-wide consultations.

Ex-ante policy evaluations are also fundamental to understand, before implementing, the potential impacts of the solutions proposed. In general, technicians and planners make use of quantitative methods and simulation models for economic and technical analysis. Simulation models can be used to forecast ex-ante and also validate ex-post the impacts of the solutions implemented, mostly from a technical point of view [13]. While quantitative methods to evaluate ex-ante the effects of transport policies from a technical/economic point of view are well-established [14], in general there is a lack of methods aimed at assessing policy acceptability from a stakeholder’s point of view. From a modelling standpoint, individual interests and the ensuing behavioural changes that a modification of the status quo would most likely induce, must be included in the framework used when dealing with policies influencing the overall city logistics framework. Freight behaviour research aims at understanding agents’ behaviour and exploring their preferences with respect to alternative system configurations [15–17]. Empirical survey data are necessary to estimate behavioural models capable of forecasting stakeholders’ reactions to alternative policies based on agents’ stated preferences [18–21]. Nevertheless, for a successful participatory decision-making process, it is important not only to forecast stakeholders’ reaction to policy change, but also to predict the outcome of an interaction process aimed at consensus building. In other words, in order to test the potential policy acceptability, it is important to model the stakeholders’ interacting behaviour. Therefore, new additional phases and tools are needed to support an effective and efficient participatory decision-making process. This is even more relevant for UFT policy-making, considered its peculiarities compared to passenger transport [22], i.e. the: (1) strong influence of the private sector, characterized by heterogeneous stakeholders with conflicting interests [11], (2) lack of knowledge and understanding of UFT problems [4], (3) relatively new inclusion of freight transport in the sustainable mobility planning agenda [23]. To this purpose, the authors propose an integrated extension of the decision-making model described in [24], which consists of three parallel and intertwined processes, combining a (i) cognitive rational approach to organising the decision-making process, (ii) a five-level stakeholder engagement process, and (iii) quantitative analyses and methods. The “fourth leg” introduced consists of behavioural analyses based on sound data and models aimed at supporting the stakeholder engagement process and, thus, the overall decision-making process. These new analyses complement the technical ones aimed at identifying and evaluating effective and shared policies, by considering both their feasibility and acceptability (Fig. 1).

Under this respect, the paper addresses the problem of complex multi-stakeholder UFT policy-making from a public authority perspective, by proposing a procedure aimed at a stakeholder behavioural analysis based on integrated models.

The remainder of the paper is organized as follows: section 2 provides a literature review of some useful methods for participatory decision-support, with the intent of contextualizing and introducing the innovative modelling approach; section 3 presents the participatory decision-support procedure based on integrated modelling approaches; section 4 describes an application of the methodology in a case study; section 5 shows the implications for UFT planning; section 6 draws some conclusions and discusses future research.

Even if public participation is considered important for the success of a decision-making process, in general it is not well-structured and there is a lack of group decision-support methods (GDSMs) that can guide stakeholders’ involvement towards well-thought-out decisions. In this respect, the gap of knowledge between methods used for technical/economic evaluation and those exploited to support stakeholder engagement is high. The rationale of this section is to provide a broad, yet not exhaustive, review of some interesting methods used for participatory decision-support, both in passenger and freight transport planning and in generic decision-making process oriented to stakeholder involvement. The choice of including non-specific freight transport tools is the result of a lack of knowledge and research in the field of transport, in general, and freight transport, in particular. Some of them are based on well-known decision-support methods and tools (i.e. focus groups, multi-criteria decision-making methods, geographic information system (GIS) and discrete choice models) while others are more recent and involve social analysis (i.e. social network analysis and agent-based modelling). Some of them help stakeholders understanding the problem and making decisions (i.e. multi-criteria decision-making methods, Participatory GIS, focus groups), while others support practitioners and policy-makers by providing insight into stakeholders’ preferences and complex social aspects (social network analysis, discrete choice models, agent-based modelling). In this respect, the former can be considered as decision-support tools for stakeholder consultation and participation, while the latter can provide support for stakeholder analysis and forecast of their choices and preferences.

Focus groups (or Mega Focus groups) are among the typical methods that can be used to support a participatory process [25]. They are meetings made up of a selected group of people (bigger in Mega Focus Groups) that can provide qualitative data and information thanks to the participation in a discussion focused on a given subject [26]. In general, a moderator leads the group and guides the conversation with the aim of collecting “rich, detailed data” [27]. They are particularly useful in an initial phase of a decision-making process to perform a preliminary analysis and pave the way to further in-depth stakeholders’ preferences investigations. In this respect, they can provide the input necessary to design a stated preference survey [3, 28–30]. In UFT, they are useful to gain insight on freight distribution problems. For example, one can organize separate stakeholders’ focus groups first to get single participants’ informations and then a joint meeting where participants can discuss together and find a common definition of the problem nature [3]. A learning process based on sequential focus groups can help adjusting initial opinion differences thanks to a dynamic dialogue among participants [31].

Multi-criteria decision-making (MCDM) methods are widely used to evaluate transport solutions from a multi objective multi stakeholder point of view. With MCDM methods alternative comparison is based on the assessment of their contributions to different evaluation criteria, that can be expressed by heterogeneous measures (monetary, physical and linguistic). In transport planning, multi-criteria analysis has been widely used to overcome the limits of traditional Cost-Benefit Analysis (CBA). CBA postulates the choice of the project generating the highest increase in the net utility of the reference community, among a set of competing alternatives. Annema [32] argues if CBA or MCDM is the appropriate tool and proposes an integrated approach based on a trade-off sheet which embraces them by adding a post-analysis phase of political debate and voting. In [33] a review of more than two hundred publications concerning the use of MCDM for transport projects evaluation is provided, defined as “a set of possible human activities that organize, optimize or facilitate the movement of persons or freight from location a to location b”. The authors acknowledge from their analysis that in most of the current MCDM analysis of transport projects, stakeholder participation is not appropriately arranged, while only in a few cases it is considered crucial and included in the whole analysis via a well-structured procedure. In this respect, Macharis [34] proposes a method, called Multi Actor Multi-Criteria Analysis (MAMCA) that, starting from single-stakeholder analyses, evaluates the different alternatives through an overall MCA to derive stakeholder-driven priority rankings among different alternative projects. The MAMCA method was used in different contexts of transport decisions to assess sustainability judgments about alternatives in transport projects (e.g. [35–37]) and to derive a framework for city freight distribution (CD-MAMCA) [38]. UFT solutions have been largely explored with the support of different MCDM methods and hybrid approaches [39–42]. Among the variety of MCDM methods, the Analytic Hierarchy Process (AHP) by Thomas Saaty [43] has proved to be suitable for complex multi-stakeholder multi-criteria decision-making [44–47]. It is a process based on the decomposition of a decision-making problem into a tree structured decisions’ hierarchy and pairwise comparisons through the building of matrixes to derive priority scales and weights. The priority of each stakeholder can be aggregated according to different methods resulting in a group priority ranking [48]. Le Pira et al. [47] demonstrate that a combination of AHP with consensus building procedures (i.e. Delphi method) is useful to support complex multi-actor decision-making processes.

MCDM in transport can largely benefit from the support of Geographic Information System (GIS), due to the intrinsic spatial nature of transport systems and the capability of GIS maps to easily visualize the impacts of transport choices on land use, environment and communities [49]. In this context, Public Participation GIS (PPGIS) or Participatory GIS [50–53] have been developed as a powerful tool for supporting non-experts’ involvement in transport decision-making processes, thanks to the power of visualization which increases the awareness about the decision to be made. According to Piantanakulchai and Saengkhao [44], GIS-based transport models combine Engineering Model (i.e. mathematical model that relates physical quantity regarding the impact being considered in space) and Weight Decision Model (i.e. model that relates physical quantity in engineering model with social preference). The spatial information in GIS and the result of objective weights can help stakeholders’ participation in the decision-making process.

Having a clear insight on the actors involved in the decision-making process and the interactions among them is helpful to sort out the tools that can be used to support practitioners and policy-makers in understanding the dynamics of participatory processes and in representing stakeholders in the relevant social networks. In this respect, Social Network Analysis (SNA) allows to quantify the social importance of a given individual in a network via centrality indexes and to understand the potential problems deriving from topology [54]. Social networks are graphs consisting of nodes (i.e. the social agents) and links (i.e. the relationships among them). They fall within the category of complex networks, whose structure is irregular, complex and dynamically evolving in time [55] and adequate methods are needed to study their structure and dynamics. The use of SNA in the field of stakeholder engagement can simply consists of stakeholder mapping or it can include centrality measures [56, 57]. Social networks re-creation can be performed using different methods, e.g. the “name generator” technique, which identifies social networks through in-depth interviews [58, 59], or the “snowballing” technique, where a small number of people are asked to nominate others, the nominees are asked for further nominations and the network builds up like a snowball [54]. There are also automatic tools which can create a network and extract information from it, such as UCINET [60] or StakeSource, a web-based tool that uses social networks, a “crowdsourcing” approach to identify and prioritise stakeholders and their requirements [61].

If SNA is useful to understand the social relations and importance of stakeholders, ex-ante behavioural analyses are important to provide insights into stakeholders’ behaviour and preferences. Discrete choice models (DCM) aim at analysing the behaviour of a decision-maker when choosing among different (discrete) alternatives, assuming that she maximizes her utility (rational decision-maker). They can be used to investigate stakeholders’ preference heterogeneity in order to forecast their individual choice behaviour related to policy-making, i.e. as a tool for stakeholder’s behaviour analysis and forecast. Economic analysis of individual discrete choices makes use of random utility theory [62, 63]. Stated preference (SP) experiments are useful as a basis for DCM to study policy acceptance in terms of stakeholders’ reaction to policy change. A choice experiment aims at acquiring high quality data to generate reliable and useful estimates of the parameters of interest, with different response format (i.e. choice, ranking or rating) [64]. There are many studies across different sectors that relate community and stakeholder acceptance of public policies with discrete choice theory, as a way to facilitate improved community (or stakeholder) analysis [65–70]. DCM have been used to determine behaviourally consistent policy evaluation in the field of UFT policy-making (e.g. [20, 21, 71, 72]). In this respect, the authors acknowledge that stakeholder-specific data acquisition is needed not only when local authorities want to adopt distinct policy instruments for different stakeholders or stakeholders types, but also when homogeneous policies impacting the various stakeholders are considered. Other relevant application of SP data and DCM for analysing stakeholders’ behaviour with respect to UFT policies can be found in [15, 73–78]. DCM are useful in helping decision-makers defining the possible policies stakeholders most prefer, given their stated choices for alternative policy configurations. The main advantage is the possibility to estimate ex-ante the willingness to pay [79] or the substitution rate between different policy characteristics (e.g. access fee and number of loading and unloading bays inside a Limited Traffic Zone). Although their significant contribution to increasing the knowledge of stakeholders’ preferences based on a sound microeconomic analysis, DCM are not well suited to investigate dynamic interactions among actors that should, on the contrary, need to be addressed to provide effective policies [22].

Agent-based models (ABM) are typically used to model complex systems and reproduce communities of autonomous and intelligent agents, acting and interacting with the environment and the other agents according to their interests [80]. One of the main ABM characteristics is the emergence of collective phenomena, not easily predictable from agents’ simple behavioural rules. In the field of freight transport, they have been traditionally used to simulate stakeholders’ interaction and trigger their reaction to some policies or regulations [81–86]. In the field of consensus building and collective policy-making, ABM are well suited to reproduce the dynamic interactions occurring within communities of stakeholders, linked in social networks and cooperating to find a convergence of opinions towards a shared solution [87]. Le Pira et al. [25, 88–90] reproduce participatory processes in passenger and freight transport planning via opinion dynamics models [91], characterising the agents with opinions and with an interactive behavioural rule aimed at consensus building. The models proposed allow to reproduce different contexts of decision-making processes involving stakeholders, understanding the role of network topology and other sensitive variables in reaching a convergence of opinions and avoiding the risk of decision deadlock due to inconsistency. The output of the simulations can be used to understand which conditions favour the convergence of opinions among stakeholders and which policy packages should be considered as good candidates that will likely be accepted by stakeholders.

In this context, DCM play a fundamental role in characterizing agents with their own utility functions based on data derived from stated preference exercises. In fact, one of the main ABM limitations is the lack of reliable data to feed the model that is trying to mimic agents’ behaviour. The DCM-derived utility functions represent a valuable input for ABM since they describe agent-specific preferences for alternative policy packages [22].

This section aims at illustrating, through a case study, how to apply the participatory decision-support procedure for UFT policy-making. The case study refers to urban freight distribution in Rome, where the extensive work performed in [17, 21, 64, 72, 97] provides an in-depth knowledge of the decision-making context and of the preferences of heterogeneous stakeholders. In this respect, the authors demonstrate, using the case of Rome as reference, how an agent-specific approach, both when acquiring and modelling SP data, is needed to: 1) deal with heterogeneity in preferences, 2) increase decision-makers’ awareness and 3) help taking better decisions [17, 21, 72]. In [17] transport providers’ preferences for alternative loading bays and pricing policies are investigated and results underline the relevance of both preference heterogeneity and non-linear attribute effects. Gatta and Marcucci [21] test the implications deriving from the presence of inter-agent heterogeneity (between transport providers, retailers and own-account operators) obtaining an “optimal” policy composition of an improving and equally impacting intervention on all agent-types’ utility based on heterogeneous willingness to pay considerations. Gatta and Marcucci [72] shows the potential policy distortions and bias if a stakeholder-generic approach is used instead of adopting a well-defined stakeholder-specific perspective in both data acquisition and modelling.

Besides, an application of the integrated ABM + DCM approach related to Rome’s case study has been performed in [92], by implementing an ABM fed with DCM agent-specific utility functions, proving its usefulness in obtaining a ranking of plausible policies that maximize consensus building while minimizing utility losses thanks to the negotiation process.

The proposed modelling approach produces an added value for UFT policy-making and it can be appreciated in an UFT planning context. In fact, along with technical and economic analyses, the stakeholder behavioural analysis contributes to the ex-ante policy assessment needed to support policy-makers in taking well-thought-out decisions.

Cascetta et al. [24] underline the importance of including stakeholders’ engagement activities (green blocks in Fig. 4) running in parallel to the traditional decision-making process (blue blocks), supported by quantitative analysis and modelling (yellow blocks). This “three legs” planning framework aims at building an effective communication strategy fostering the emergence of a stakeholders’ coalition supporting the plan rather than an actual collective decision-making. As reported in section 1, a decision-making process for UFT planning requires a deeper knowledge of the heterogeneous nature and preferences of the stakeholders involved, whose aims and behaviours can be hardly predicted by conventional transport models. Actually, stakeholder behavioural analysis can and should be considered as an additional set of activities (red blocks) representing the “fourth leg” of the UFT planning process (Fig. 4).

The “fourth leg” introduced consists of a: (i) preliminary analysis aimed at defining the components of the alternatives to be submitted to stakeholder evaluation, which could benefit from the feedbacks of phase (iii); (ii) survey aimed at eliciting stakeholder preferences – via SP exercises – and understanding the nature of their existing social interaction; (iii) simulation tool integrating DCM and ABM to provide (iv) scenario simulations of interaction processes among stakeholders with the aim of unveiling acceptable alternative packages.

The stakeholder behavioural analysis proposed is strictly linked with the overall decision-making process, since the technical/economic feasibility of the “most acceptable policy package” derived from the simulations has to be evaluated by quantitative methods and tools, capable of simulating the effects of alternative transport system’s configurations¹ (i.e. transportation system analysis of Fig. 4). Finally, assessment results are presented to decision-makers and stakeholders, so to support the final decision that should be consistent with the previously identified objectives (see Fig. 4).

The modelling framework proposed within a UFT context can realistically be applied to any decision-making process dealing with transport planning, especially when the complexity of the decision, the heterogeneity of the actors involved and the interaction of mixed competitive and cooperative behaviour are relevant for the implementation and the success of the plan, as in the case of UFT.

This paper presented a procedure based on an integrated modelling approach to support stakeholder involvement in the decision-making processes regarding urban freight transport solutions. The integrated modelling approach is based on a well-thought-out integration of discrete choice models and agent-based models to take into account stakeholders’ heterogeneous preferences and simulate their interactive behaviour in a consensus building process, providing useful suggestions for policy-makers on the potential acceptability of a set of policies to be eventually discussed with stakeholders. The integrated modelling approach is able to combine the advantages of the two methods while overcoming their weaknesses, since it is well grounded on sound microeconomic theory providing a detailed (static) stakeholders’ behavioural knowledge, but also capable of reproducing agents’ (dynamic) interaction during the decision-making process. It produces an added value for UFT policy-making and it can be framed in the overall context of transport planning. In fact, together with technical and economic analyses, the stakeholder behavioural analysis proposed contributes to the ex-ante policy assessment needed to support decision-makers in taking well-thought-out decisions. In future development of this research, the authors will test the procedure in different decision-making contexts regarding urban freight transport solutions, trying to adapt it to the specific problem type. In this sense, the behavioural analysis proposed can be considered as the general framework for a participatory decision-support tool.