Adaptive decision making in multi-stakeholder retail planning

I.I. Janssen

Research output: ThesisPhd Thesis 1 (Research TU/e / Graduation TU/e)

1192 Downloads (Pure)


The decision where to locate new retail facilities is increasingly more a multi-stakeholder decision instead of a single-actor decision. In the past, the Dutch Government had a strong hand in determining the program and location for new shopping centres. Since the introduction of the newest national policy document on spatial planning, the so called "Nota Ruimte", the Dutch government decided to relax restrictions for all retail categories to be located at peripheral locations. Furthermore, with this new policy the responsibility for planning decisions was delegated to local governments. This change in retail planning policy gave room to real estate developers (often cooperating with retail firms) to initiate peripheral retail development. Thus at the present time, planners, retailers and developers, as main actors involved in retail planning, meet each other at the local policy level, to decide on the location of new retail developments. Previous studies assumed that stakeholders involved in location decisions make independent, sequential decisions. However, since location decisions are made in a larger context of stakeholder interactions, stakeholders have become jointly responsible for the full development process. Nevertheless, the decision behaviour of these stakeholders is neglected within location decision literature. The aim of this study is: (i) to reveal preferences for retail location options of different stakeholder groups, and (ii) to measure the degree in which these preferences are influenced by the preferences of other stakeholders. For this second purpose, the concept of adaptive behaviour is defined as the phenomenon that a decision maker adjusts his/her preferences in accordance with the preferences of other decision makers in order to reach consensus. It is supposed that adaptive behaviour plays a role since stakeholders interact during negotiations and influence each other’s viewpoints in order to reach agreement. To meet both purposes an experimental research approach was used. Three stakeholder groups (local governments, developers and retailers) were invited to participate in a Web-based conjoint choice experiment. Before discussing the experiment, the main part of this study, the first chapters of this thesis give a theoretical foundation of the retail planning problem, the retail planning process and multi-actor decision making in general. Chapter 2 of this thesis starts with an explanation of the retail location decision problem. It discusses how retail planning policy in the Netherlands has shifted from a restrictive to a relaxed policy that allows new retail developments to be located at peripheral locations. It also shows that decentralization of planning responsibilities is coherent with more general shifts in planning views. Furthermore, it argues that retail planning decisions have to be made in a very dynamic, constantly changing, decision environment. Consumer behaviour is changing fast, leisure has become more important (often combined with shopping), and the importance of internet as retail channel has increased. On the supply side, increases of scale and internationalization of retail firms have changed the demand for retail property. Finally, market dynamics had a big influence on investor behaviour. Although the interest from institutional investors in shopping centres as an investment asset grew, the financial crisis led to a slowdown of investment and development activities. Chapter 3 takes a closer look at the processes regarding retail planning decisions from both the planner’s and developer’s perspective. To understand the course of decision making, insight in formal planning procedures and legal instruments is needed. The most important aspect in this perspective is that, since retail planning policies have been decentralized and relaxed, local and regional governments are now in the middle of revising their structure plans with retail being part of it. In the meantime, several new peripheral retail developments have been initiated which anticipate deregulation. Three of these initiatives were discussed in this chapter. These case studies showed the importance of agreement about the best retail structure for a particular area to get new peripheral retail plans that fit this approved retail structure. Missing consensus will certainly lead to frustration during the development process. Moreover, the case studies showed that even if private and public stakeholders jointly agreed on a plan proposal, the political decision making process that follows may frustrate plan development. Subsequently, in Chapter 4 the concept of multi-actor decision making is discussed in more detail. It explains that negotiations on retail plan proposals ought to be joint (cooperative) decisions although decision entities may also show non-cooperative behaviour. During the negotiation process the preferences of each decision maker may be influenced by interactions with other decision makers. When showing cooperative behaviour stakeholders adjust their preferences to reach consensus. This adaptive behaviour is the focus of this study. Preferences can also be influenced by interactions that are not preceded by real actions between the negotiators, such as interactions by media, other stakeholders that are not the negotiators (like pressure groups), or experiences with former interactions. To explain why decision makers may adapt their preferences, different reasons are discussed based on a literature review. These reasons include differences in power positions, interests and perceptions. Choice modelling is proposed to be a suitable approach to measure retail location preferences and adaptive behaviour of stakeholders. It can deal with multiple discrete attributes and can be applied to multi-stakeholder settings. In the next part of the thesis, the research approach to measure preferences and adaptive behaviour, the data-collection and data-analysis are discussed. In Chapter 5 it is first argued that traditional conjoint experiments and choice modelling techniques have to be extended in order to collect data on adaptive behaviour. It is explained how adaptation and context variables can be derived from the data collected with choice experiments. Each respondent had to choose the most preferred alternative (retail plan) from sets of alternative retail plans. The experiment consists of two parts. In contrast to the first part of the experiment, in the second part the preferences of the other stakeholders were added to the decision context. The context variables could be derived by measuring the differences in preferences between the first and the second part of the experiment. The adaptation variables were obtained by assuming positive additional attribute effects for the alternatives chosen by the other stakeholders and negative attribute effects for the other choice alternatives. These adaptation variables measure the degree that preferences of decision makers are affected by the preferences of other the stakeholders. Based on these principles, a Web-based conjoint choice experiment is developed as described in Chapter 6. Three stakeholder groups (local governments, developers and retailers) were invited to participate in a Web-based conjoint adaptation experiment. Stakeholders were asked which retail plan alternative they prefer for expanding the existing retail structure of the imaginary city "Shop City". Expansion of retail supply was possible in three retail categories (Toys and Sporting Goods, Home Electronics & Media, Fashion) and a Restaurant. These four categories represented the attributes of the experimental design. The levels for each category represented possible locations for expansion of these retail categories; 1) adjacent to a sport stadium, 2) an expansion of a furniture strip and 3) the inner city. The choice options reflect typical current retail developments, in nature and size, in the Netherlands. The data-collection procedure and the response have been discussed in Chapter 7. Every respondent had to make two times fifteen choices. Tests with students showed that this number of choices was good to handle without becoming indifferent. A total number of 170 respondents (67 developers, 67 local governments and 36 retailers) made 5100 choices, which was a good base for model estimations. Although the number of retailers that took part in the experiment was relatively small, it reaches the number that was aimed for, resulting in satisfying modelling results. Based on the characteristics of the respondents we can conclude that they were a good representation of the stakeholder groups in practice. The analysis of the results that are discussed in Chapter 8 provides interesting insights that help to explain retail planning decisions. Results show that adaptive behaviour plays an important role in these decisions. Tests showed that for all three stakeholder groups the models including adaptation variables performed significantly better than the models without adaptation variables. Except for some developer’s models, including context variables did not lead to significantly better model estimations In general, the results reflect the background of the stakeholders. The group of developers appeared to be the most adaptive. Developers facilitate with their development plans market demand and are willing to adapt their viewpoint to the opinion of the other stakeholders. Retailers turned out to be the most persistent in their viewpoints. They represent market demand and as such do not allow their preferences to be influenced by choices of developers nor local governments. Finally, local governments behave somewhere in between. Almost all adaptive behaviour that was estimated was cooperative, implicating that in general stakeholders are willing to positively adapt their viewpoint to the preferences of others. In general, negative part-worth utilities did not turn positive, but became less extreme, potentially increasing the level of consensus. Regarding the preferences of the different stakeholders concerning the different retail plan options it can be concluded that these are typical for the current Dutch retail market. All stakeholder groups believed that Fashion should not be located at a peripheral retail location. This suits with the general opinion that buying clothes is considered to be a recreational shopping activity and for that reason, Fashion should be located in the inner city shopping areas. Peripheral locations are regarded to be the location for goal-oriented shopping motives. With respect to the location of the retail categories Toys & Sporting Goods and a Restaurant, stakeholders appeared to be rather indifferent. For the developers, however, the most preferable option for locating the Restaurant was the furniture strip. All stakeholders reject the option of locating the retail category Home Electronics & Media near a sports stadium. The group of developers prefer to locate this retail category at the furniture strip while the retailers and local governments are indifferent in locating this retail category at the furniture strip or the inner city. In general, it seems that although retail planning legislations in the Netherlands has been relaxed, public as well as private stakeholders still show conservative opinions in locating non-food branches other than furniture out of the city centre. Including adaptation variables significantly improved the model estimations. The models’ performances could even be improved more by taking heterogeneity in consideration. It was found that Mixed Logit models, taking into account taste differences, performed significantly better than standard MNL models, Furthermore, models including variables indentifying subsamples performed significantly better than models without. To conclude, this study showed that extending traditional choice experiment enables researchers to capture behavioural aspects like adaptive behaviour within multi-stakeholder decision models. The results of the experiment do not explain why stakeholders adjusted their preferences, although a literature review gave some suggestions. Future research could explore these motives underlying adaptive behaviour by applying in-depth interviews, for example. Finally, future research could focus on the way information about preferences and adaptive behaviour in retail planning can be used as input for new location decision models that cover several negotiation steps.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Department of the Built Environment
  • Timmermans, Harry J.P., Promotor
  • Schaefer, Wim, Promotor
Award date26 Apr 2011
Place of PublicationEindhoven
Print ISBNs978-90-6814-636-3
Publication statusPublished - 2011

Bibliographical note


Fingerprint Dive into the research topics of 'Adaptive decision making in multi-stakeholder retail planning'. Together they form a unique fingerprint.

Cite this