Wander Jager University of Groningen (NL) Dept. of Marketing

1. Simulating consumer behavioura perspective on how to model consumerbehaviour for environmental policy Wander Jager University of Groningen (NL) Dept. of Marketing

2. Environmental relevant behaviour • Environmental problem awareness plays at best a limited role • Finances and daily hassles play a dominant role in deciding to behave environmentally friendly

3. Environmental (relevant) behaviour • We have to realise that environmental outcomes often hardly affect the decision-making of consumers. • Consumers may behave environmental friendly without environmental concern playing a (relevant) role as driver of that behaviour. • Labelling behaviour as ‘environmentally relevant’ does not provide additional understanding of the drivers of that behaviour, nor does it provide extra cues for effective policy measures. • Hence, in analysing and changing environmental relevant behaviour, standard behavioural analysis applies.

4. Behavioural analysis • Needs as motives of consumer behaviour • Consumer decision making: cognitive effort & individual and social dimensions

5. Human needs: Maslow (1954)

6. Human needs: Max-Neef (1992) • Subsistence • Protection • Affection • Understanding • Participation • Leisure • Creation • Identity • Freedom

7. Human needs • Lower needs (related to daily hassles) are experienced on a more frequent basis and more directly than higher needs • Higher needs, especially related to the environment, are more cognitive (not directly and personally experienced), and are more likely to dominate cognitive responses (questionnaire) • Lower needs are more emotionally driven, and may have a strong effect on behaviour without cognitively being experienced.

8. Human needs • Behaviour may have effect on different needs simultaneously: • destroyers or violators • pseudo-satisfiers • inhibiting satisfiers • singular satisfiers • synergistic satisfiers

9. Human decision-making • Cognitive effort • Social versus individual orientation

10. Decision-making People have limited cognitive resources. The allocation of our limited cognitive resources depends on a.o: • Consumer involvement (relevance of needs) • Complexity of the decision • Newness of the decision • Constraints (e.g. time, cognitive capacity) Low involvement will result in the use of simple strategies using limited amounts of information.

11. Decision-making Using social information is often a smart strategy to save on cognitive effort. Social decision-making is more likely when: • Lack of clear information (complex, social relevant) • Visibility of behaviour is high or communication is easy • Higher social susceptibility (personality)

12. Decision-making • Exchange of information is more likely if people are similar on the relevant dimension of comparison - Network effects play a pivotal role

13. Why multi-agent simulation? Complexity Many interacting people/consumers Heterogeneity of population Easy experimentation with policy measures

14. GLM versus social simulation SocSim: what are more and less likely routes? SocSim: How to respond to different (unanticipated) developments? SocSim: How to sail? GLM: where do we arrive given the current situation? GLM: How do changes in the instrumentarium affect the expected arrival? GLM: Where do we arrive?

15. What behaviour to simulate? • Environmental behaviour does not exist • We should model consumer behaviour in general

16. Why not to use too simple rules • Simple rules – e.g., as used in econophysics – do not represent the behavioural drivers and dynamics that typify consumer behaviour • Many policy measures are aimed at changing behavioural drivers and processes • A simple (too) model may mimick the real system (calibration), but that does not guarantee that testing ploicy measures in such a context lead to realistic behavioural changes

17. What to include in agent rules? • Human needs • Human decision-making

18. Example • Lakeland

19. Goals of market simulation • Understand market dynamics emerging from individual consumer behaviour (micro-macro dynamics) • Explore the effects of various interventions/producer behaviour on these dynamics

20. Market dynamics and social simulation • Several simulation models have been developed to model market dynamics • These models generally lack an integrative perspective on marketing strategies • Marketeers (and policy makers in general) would benefit from having simulation tools that also allow for testing marketing strategies

21. Market dynamics and social simulation • Agent architecture has to reflect the key-drivers of consumer behaviour • Marketing strategies can be implemented as affecting these key drivers

22. The four P’s of Marketing (McCarthy, 1960) product pricing placement promotion

23. Product • brand name • functionality • styling • quality • safety • packaging • repairs and support • warranty • accessories • Services

24. Price • pricing strategy (skimming, penetration) • retail price • volume discounts and wholesale pricing • cash and early payment discounts • seasonal pricing • bundling • price flexibility and price discrimination

25. Placement • use of distribution channels • market coverage • specific channel members • inventory management • warehousing • distribution centres • order processing • transportation and reverse logistics

26. Promotion • promotional strategies (push & pull) • advertising • personal selling & sales force • sales promotions • public relations and publicity • the marketing communications budget

27. Formalising the four P’s • Formalising all the factors just mentioned would – if possible – result in a model not accessible for research • A simplification is required for developing a transparent simulation model • The formalisations as presented are meant as a framework, simpler models can (and should) be used, and if necessary extended using the framework

28. Product • Product characteristics relate to individual preferences • Vector model of preferences: the more, the better (quality, service) Uinj = Ajn With: Uinj = Utility of consumer i on attribute n for product j Ajn = Score of product j for attribute n

29. Product • Ideal point model of preferences: relative position on a scale(design, colour, taste) Uinj = 1 - |Ajn - Pin| With: Uinj = Utility of consumer i on attribute n for product j Ajn = Score of product j for attribute n Pin = Preference of consumer i for attribute n

30. Product • Besides individual preferences, consumers also have social preferences for products • Networks play a critical role in social effects Uinj = Nj/N With: Uinj = Utility of consumer i on attribute n (here the social attribute) for product j Nj = Number of neighbours consuming product j N = Number of neighbours

31. Product • The utilities (vector, ideal point and social) are summed to constuct a total utility. Beta indicates the relative weight of each utility in the total utility With: Uij = Utility of consumer i for product j, ranging from 0 to 1 ßn = Weighting of attribute n, ranging from 0 to 1 Uijn = Utility of consumer i for product j for attribute n

32. Product • The weighting of utilities can be different for different agents, thus including heterogeneity in the consumer population With: Uij = Utility of consumer i for product j, ranging from 0 to 1 ßn = Weighting of attribute n, ranging from 0 to 1 Uijn = Utility of consumer i for product j for attribute n

33. Product • The decision process of consumers can be represented by the values of the betas • Cognitive effort: the number of product aspects taken into account (involvement) • Social v.s individual orientation: weighting of social utility

34. Product • Note! The formulation of utility has the lay-out of a regression formula • But: for each simulated consumer this formula may be different. Moreover, the utilities and their weigting are subject to change

35. Price • The concept of value-for-money is being used to link price to utility • the value for money will be closer to the utility of the product the lower its price and the higher the consumers budget With: Vij = Value for money of product j for consumer i Uij = Utility of consumer i for product j Pj = Price of product j, ranging from 0 to 1 Bi = Budget of consumer I, ranging from 0 to 1

36. Placement Focus on distance: - Simple formalization: distance as additional attribute in the model - Heterogeneity in distance score expresses the distance to a buying location. - Weighting the distance attribute (with a ß) distinguishes between markets where distance is important (e.g., groceries) versus unimportant (e.g., e-commerce)

37. Promotion – by producer • Convince consumers to attach more weight to a product attribute on which the product scores well (increasing the ß) • Convincing the consumers that their utility for attribute n would be higher than they currently believe (increase Uinj). • Inform consumers about other consumers (famous role models) that already use a product, thus affecting the social attribute.

38. Promotion – by producer • Who to address? • Mass media • Random consumers • Consumers with particular characteristics (segments) • Clusters of connected consumers • A mix of strategies?

39. Promotion – consumers: word-of-mouth - Specific information: exchange information concerning the product utilities (Uinj), e.g. fuel consumption of a car - Generic information: discuss the importance of certain attributes (weighting of attributes), e.g. the importance of safety of a car. - Norms: consider the number of neighbours consuming a particular product without considering further information (social attribute as defined in product)

40. Promotion – consumers: word-of-mouth -Network effects are critical Regular lattice Small World Scale Free

41. Empirical data • Macro level sales data (& timing of marketing strategies) – development of market shares over time and indication of marketing effects • Micro level sales data (loyalty card data) – how do consumers behave in a market • Micro level data on decision-making – getting grip on the decision-making process of consumers (attributes & weights)

42. Conclusions • Formalising the four P’s provides a perspective on: • modelling complexities in markets • implementing and testing marketing strategies in complex markets • The linkage of simulation models to emperical data on both the micro and macro level

43. Research organisation formalisation Micro level data (networks) Agent & product characteristics hypothesising experimentation Macro level data (markets) validation Simulation of market dynamics