Selective consumers preference and consumption rationality in market for preferred good : a case of film festival

SELECTIVE CONSUMERS PREFERENCE AND CONSUMPTION RATIONALITY IN MARKET FOR PREFERRED GOOD:

A CASE OF FILM FESTIVAL

This research investigates consumer’s consumption by extending preference as an attribute of rationality. While previous studies claim that consumers are often irrational in their consumption behaviour, this study proposes that it is not the case. By using the concept of selective consumer – those who have strong preference towards certain goods – this research finds that in film festival setting where the access for preferred good is abundant, selective consumer utilises their preference and subsequently consumes more. This research also shows that previous consumption in similar market setting holds little significance for selective consumer’s present consumption and that there is a curvilinear relationship between selective consumer’s consumption and geographical distance to the market. This research contributes to the consumer rationality, preference and consumption behaviour discourse, by showing that when their preference is incorporated to the context of their consumption, consumer are rational in their consumption behaviour.

M.C.M. Silalahi (10621652) First Supervisor: Angelo Tomaselli

Second Supervisor: Joris Ebbers Master Thesis

MSc Business Administration

Entrepreneurship and Management in Creative Industries Track

Statement of Originality

This document is written by M.C.M. Silalahi who declares to take full responsibility for the content of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Acknowledgement

This thesis would not be possible without the assistance of many people, therefore I would like to thank the following:

Angelo Tomaselli, my supervisor.

Cinemasia Team, for giving me the opportunity to be a part of a very pleasant festival, in particular Bianca Kuijper, Lorna Tee, Yuen Kwan Lo and Esther Schmidt for helping me out with this research.

Cineville, for their kind cooperation, especially Thomas Hosman who had given me first-hand access to reliable and sizeable dataset.

On a more personal note:

My late father, Dr. Pande Radja Silalahi, who taught me so much by doing so little; and whose sudden death at the beginning of my master study has assured me that the world would never cut me any slacks.

My extended family, whom I can always rely on for added pressure.

Table of Content

1. Introduction ……… 4

2. Theoretical Background ……… 6

2.1. Consumer Rationality ………....… 6

2.2. Consumer Preference ……… 8

2.3. Consumption and Selective Consumer ………. 9

3. Hypothesis Development………... 11

3.1. Product Categorisation …………...………. 12

3.2. Segmentation ………... 14

3.3. Conceptual Model ………...… 17

4. Empirical Setting………... 18

4.1. Creative Industry and Cultural Goods ………. 18

4.2. Film Industry ………... 20

4.3. Art Film Audience as Selective Consumer ………. 21

4.4. Dutch Film Market and Unlimited Subscription Card ……… 21

4.5. Film Festival ……… 22

5. Methodology ………. 23

5.1. Card Subscription – Cineville ………. 24

5.2. Festival – Cinemasia ………... 25

5.3. Sample and Procedure ………. 26

5.4. Variables and Measures ………..… 28

6. Result ………...………….. 30 6.1 Descriptive Statistics ……… 30 6.2. Correlation ………... 31 6.3. Hypotheses Testing ………. 34 7. Discussion………..………. 39 7.1. Contribution ……….. ………. 39 7.2. Practical Implication ………... 41

7.3. Limitation, Future Research ………..……….. 42

8. Conclusion………..………...……. 43

9. References………..………...……. 45

1. Introduction

David Hare (1995, p.15) wrote in one of his plays “We never do anything because we

might actually enjoy it. We do it so we can write it all down”. Anecdotally, this quote seems

to sum up people’s behaviour in today’s contemporary world. Entangled with the prominence of social circle and the ever connected society, the world seem to diverge into a more singular society where people do things to be seen doing it. This tendency, which leaves so little room for genuine personal preference, seems to also be the case when it comes to consumption.

At first glance it might seem completely irrational for someone to do or consume something just for the sake of writing it down. To illustrate, it would seem irrational at first for someone who despises academic work to write academic papers just so they can make their CV more colourful. However, enjoyment in Hare’s context is not clearly defined and therefore the act of doing something cannot be simply deemed irrational or incoherent with enjoyment. After all, it is not impossible that the act of writing itself is the actual source of enjoyment to begin with. Illustratively, it might be the case that not the actual writing of academic paper that brings enjoyment, but the colourful CV instead. Whether this rationale crosses Hare's mind while writing the play or not, is of little importance. However, this quote could be pivotal in shaping the way we understand consumer preference and how it determines rational consumption.

The notion of consumer rationality has long been debated in the field of consumer behaviour (Holbrook and Hirchsman, 1982) mainly due to the fact that rational choice plays a pivotal role in being a consumer (Bettman, Luce and Payne, 1998). Previous scholars have long associated rationality with the notion of utility maximisation (Herstein and Milnor, 1953 as cited in Shugan 2006; Becker, 1996; Zalega, 2012). However, when it comes to real practice, consumer's actions oftentimes fail to be considered rational under the theoretical standards (Graham, 1981; Della Vigna and Malmandier, 2006; Cohen and Dickens, 2002)

On the one hand, some researchers suggest that this breach against rationality is due to the fact that economic theory which examine consumer demand does not incorporate taste and preference (Potts et al., 2008). On the other hands, there are also researchers who argue that preference is not consistent with rationality tenets which include consistency (Coupey, Irwin and Payne, 1998; Bettman, Luce and Payne, 1998). In sum, prevalent literature has yet to reach an agreement on how best to bridge the discrepancy between theoretical assumption and real practices.

The abovementioned reasoning leads this research to address consumer rationality in a way that incorporate the notion of preference in order to better understand consumer behaviour. This research examines selective consumers – those with strong preference towards consuming certain goods - consumption behaviour. Using the setting of a film festival, this research introduces a market setting which offers varieties of preferred goods in order to create stronger distinction between the types of consumer. Moreover, the analyses for this setting is done through the assessment of the level of consumption by selective consumer using the proxy of screening attendance.

Together, the result of the hypotheses formulated this research will provide more insight into selective consumer behaviour and help answer the research question of “how

does a selective consumer consume in the market for preferred goods?”

The findings in this research show that in a setting where the access for preferred good is highly accessible, selective consumer utilises their preference and subsequently consumes more. This research also shows that previous consumption in similar market setting holds little significant for selective consumers’ present consumption and that there is a curvilinear relationship between selective consumer’s consumption and geographical distance to the market. By showing that consumers are rational in their behaviour when their preference is incorporated to the context of their consumption, this research counters the

academic hesitation in incorporating preferences due to its labile nature (Coupey, Irwin and Payne, 1998) and contributes to the consumer rationality, preference and consumption behaviour discourse.

This research is structured as follows: first the theoretical background will present the discourse on consumer rationality, preference and consumption, followed by the introduction of the concept selective consumers, preferred goods and market for preferred goods. Second, the hypothesis development part will delve deeper into the rationale behind the hypotheses proposed by this study. Third the empirical setting will provide a specific setting chosen for this setting and why such choice is deemed as a good fit. Fourth, the method for the analysis will be elaborated. Fifth, the result is described. Lastly, the result is discussed and the conclusion will round up this thesis.

2. Theoretical Background

In the field of consumer behaviour, the notion of consumer rationality has been long debated. Holbrook and Hirchsman (1982) go so far as saying that the study of consumer behaviour evolves from notion of rational choice. Rational choice is a crucial part of being a consumer (Bettman, Luce & Payne, 1998) and consumer rationality greatly affects the effectiveness and implementation of numerous market activities (Shugan, 2006).

Previous scholars dealing with economic theory often define consumer rationality implicitly in terms of expected utility maximization (Herstein and Milnor, 1953 as cited in Shugan 2006). Becker (1996) advances the notion by defining rationality as the quality attributed towards individual/agent whose decision is consistent, forward looking, and most importantly, utility maximising. Thus inferring that individual rationality is a function of their ability to maximise their utility. Additionally, Zalega (2012) also adds the attribute of

consistency towards rational behaviour and describes irrationality as the behaviour that contradicts consumer’s best interest and it is characterized by inherent inconsistence.

As expected from the extant definition, the notion of rationality is almost always associated with utility maximization with each different discipline apply specific attributes to the context – like for example game theory application defines rationality as the ability to take the best alternative given clear payoffs and rules (Bernheim, 1984 as cited Shugan 2006).

Nonetheless, when it comes to consumer behaviour, Graham (1981) points out that most of the decisions made in practice fail to be grasped as theoretically rational because they are only rational for the consumer. This means that in order to understand irrationality researcher needs to understand the consumer rationale as opposed to mere academic exemplification.

Indeed, academic evidences have shown that rational expectation assumed in theory can be inconsistent with consumer behaviour in practice. For example, Della Vigna and Malmandier (2006) found that consumers do not necessarily maximise their utility. Furthermore, Cohen and Dickens (2002) also state that under a specific setting, rationality based on expected utility in consumer behaviour often falls short in practice. Thus, to better understand the discourse of consumer behaviour and rationality, clear criteria for rationality in consumption needs to be established (Zalega, 2012). In another word, establishing the grounding of consumer rationality and understanding consumer’s rationale beyond academic definition is at the essence. Becker (1996) mentions that including preference as the extension of the utility maximising approach is remarkably effective in unifying varied behaviour. This brings us to the notion of consumer preference and how it plays a pivotal role in extending the rationality in consumption concept.

2.2. Consumer Preference

Understanding processes that induce consumer’s preference has long been regarded as one of the main interests of marketers and management researchers. Not only information about consumers’ preference serves various purposes from setting up price for products to developing communication strategies to market product and services; but as Gazley, Clark and Sinha (2011) point out, consumer preference works as the driver of consumption.

Insights into consumer preference are precious, especially considering that behavioural evidence has shown that consumers do not necessarily act the way economic rationality expect they would. Ben-Akiva et al. (1999) suggest that cognitive anomalies - or occurrence where individuals show significant departure from rationality - are elicited by preference. Moreover, Hoch and Loewenstein (1991) also state that there is a need to reconcile rational motives and passion in consumer research.

Potts, et al. (2008) note that economic theory of consumer demand often neglects taste and preference in favour of mere utility maximisation subject to budget constraint. Nonetheless, some studies do incorporate individual preference as one of the basic criterion of rationality; and since consumers make their choice according to their liking, individual preferences are deemed to be rational (Zalega, 2012). Ergo preference is not entirely overlooked in rationality discourse and consumers are not just governed by set of incentives but also preference.

Naturally, the justification to incorporate preference does not go uncontested. Bettman, Luce and Payne (1998) argue that oftentimes consumers do not possess an established and comprehensible preference. Coupey, Irwin and Payne (1998) also mention that a preference is often labile thus violating the consistency attribute of rationality. However, in spite of the contentions against preference as a property of consumer rationality,

Hoyer and Stockburger-Sauer (2012) argue that taste and preference are consistent and responsive to external standards, especially in hedonic consumption.

In sum, there has yet been a clear consensus on whether preference approach is the right tool to better understand the rationale behind consumer’s departure from rationality. Markusen et al. (2008) suggest that conceptual definitions deriving from the supply side often lead into confusion. This is quite problematic since researchers usually rely upon the supply side as opposed to the demand side where insights can actually be gardened through consumption pattern. Bearing this in mind, this research will try to examine consumer rationality as a function of preference through a focus on the demand side. Specifically, this research examines selective consumers, their consequent auxiliaries, and their behaviour in the setting that facilitate consistency and enable distinction between different type of consumers.

2.3. Consumption and Selective Consumer

As mentioned earlier, consumer preference drives consumption (Gazley, Clark and Sinha, 2011). Moreover, distinct preference also serves as a tool for identity building (Peltoniemi, 2015) and for consumers with stronger preference towards certain goods (dubbed in this research as preferred goods), consumption reinforces their desire for identity. Levy (1959) mentions that consumption is not merely driven by functionality but also symbolic value. People consume certain product to communicate their characteristics and to distinguish themselves as such (Belk, 1998; Holt, 1995; Solomon, 1983) and individuals express their identity by differentiating themselves in consumption. This symbolic value is in turn incorporated by behaviour which signal identity (Berger and Heath, 2007)

The perspective on preference and identity building substantiates consumer reasoning to consume more of the good they identify with. Consumer prefer products with image they

can associate with themselves. This theory is known in the consumer behaviour discourse as the 'self-congruity' hypothesis (Sirgy, 2015) and its line of reasoning leads us to the taxonomy of consumer types and their subsequent consumption. Whereas general consumers, those who do not possess or signal clear preference consume more sporadically, consumer with fixed preference takes their consumption more intensely (Adler, 1959).

Jeffres, et al. (2003) state that selectivity can be expressed on the basis of interest maximisation and the use of gratifying product. In this sense, consumers act upon their preference by actively selecting the product they consume in order to maximise their interest in lieu of utility maximisation. Bettman, Luce & Payne (1998) postulate that the more consumers behave selectively according to their values, the less vulnerable they are to information overload. Therefore, incorporating values and active selection in consumption is applicable to the preference dimension of the rationality argument. Meaning, the more consumers are selective in their consumption on the basis of the preference they hold, the more unsusceptible they become to the distraction and this translates to rationality.

Furthermore, consumers are actively in charge of what product attribute they choose to see and what they will eventually consume. Consumers are also selective in their exposure information (Tewksbury, 2005). Therefore, especially for consumer who has a well-defined preference, active selection is inevitably a part of consumption process from the get go.

Departing from the theories of consumer rationality and preference as well as consumption, this is where this research tries to fill in the gap, by examining selective consumer’s demand and consumption features. Selective consumer is defined in this research as the type of consumer with strong preference towards preferred goods and consume accordingly. In contrast to the general consumer whose preference is obscure and who are more likely to consume mainstream products, selective consumer identify themselves through their consumption of the goods they preferred, which are referred to as the preferred goods.

Having established the differences between selective and general consumer however has yet to provide a clear insight to the actual consumption of the selective consumer, which is where a situational context of a market setting where abundant access to assortment of preferred good is in chosen for this research. This specific setting will be elaborated further in the empirical setting section where the setting of festival is introduced.

3. Hypotheses Development

Insofar as the relevant literature has not shown an affirmative answer on whether rationality is a function of preference, this research argues that for selective consumer, consumption rationality stems from preference and therefore selective consumers display rational consumption in their behaviour. Meaning, this research argue that solid preference translates to rational consumption.

Nevertheless, some studies have suggested that deciding on preferred alternative does not necessarily contribute to rational choice consumption (Dhar, 1997). This is where this research tries to introduce a setting that amplifies preference, that of assortments of preferred goods. Assortment in this sense means that there are varieties of preferred goods offered in a single market.

There are different studies that examine the role of assortment in preference. On the one hand, Reibstein, Youngblood and Fromkin (1975) put forward the idea that larger assortments can lead to stronger preference. Lancaster (1990) supports this notion by pointing out that larger assortments provide opportunity to match consumer’s preference. On the other hand, Chernev (2003) states that large assortments may actually lead to the weakening of preference. Now, recalling the identity feature of consumption and consistency attribute of rationality, a set of assortments is the ideal setting to test out the effect of preference in selective consumer consumption level due to the strength of the preference. Meaning, strong

preferential identification and commitment from selective consumer neutralise the weakening attributes of product assortment.

It is often assumed that the quantity being consumed by consumer is made independently, meaning that how much consumers consume does not influence which product they choose (Simonson & Winer, 1992). However, if consumer rationality is a function of preference, providing a fixed setting with assortments of preferred goods would provide selective consumers with a platform to further exploit their maximisation process, be it utility or interest.

Moreover, Simonson (1990) also proposes that in the case where consumer is uncertain with their future preference or if they face difficulty in choosing from available alternative, their consumption becomes more varied. Having a specific setting which facilitate assortments of preferred goods roots out this problem because it provides singularity of preferred goods which in turn facilitate consumption. In this sense, the selective consumer who possesses fixed preference will not be distracted as such. Therefore, if preference does extent to rationality, then a selective consumer will consume more in a place where they have abundant access to their preferred goods. This brings us to the first hypothesis where whether selective consumer’s consumption will be strengthened or weakened in a market context where large assortment of preferred good is available is tested.

H1: In a market that offers assortments of preferred goods, there is a positive relationship between being selective consumer and level of consumption

3.1. Product Categorization

Having established the conjecture of the relationship between being selective consumer and consumption, the discussion now delves into the analysis of how selective

consumer’s consumption comes to be as such. This research propose that product categorization plays a role in consumption pattern displayed in the market where assortments of preferred good is available, since product categorization serves as a proxy to monitor product preference consistency for both the general and selective consumer. Additionally, product categorization also serves as point of reference for distinguishing selective consumption in the first place. Meaning that selective consumer preference is identifiable through their consumption of product that belongs to certain distinguishable category.

In a loose sense categorization refers to the purposeful grouping of attributes. Over the time academics have tried to dissect categorization and define categories and the process of categorization in multiple ways. For example, Kuijken, et al. (2013) defines categories as

“socially constructed cognitive orderings” or Hannan et al. (2007) who describes categories

as collective agreement based on similarities.

One's consumption patterns and tendencies to consume certain goods work as a mechanism to represent one self to others (Argo, Dahl & Manhanda, 2005; Ratner & Kahn, 2002). Chakravati and Janiszewski (2003) suggest that people’s well-established preferences lead to a limited and focused consideration set and that category labels expedite preference identification. For consumer, the presence of categories acts as identifier for different types of items or product and more than likely to help them to infer differences. This is consistent with Mogilner, Rudnick and Iyengar (2008) who state that categorization help facilitate preferential treatment as the existence of category label might signal to consumer the differences between options. Moreover, White and Dahl (2006) also state that since grouping serves as points of references on how consumers think and behave, product grouping or product categorising guides consumer's preferences. In another words, product categories serve as channels for consumer to exercise their preferences. Departing from the abovementioned reasoning, it is fitting to infer that for selective consumer, having a strong

preference might in turn lead to a more optimised consumption for the product categorised as preferred goods. Additionally, referring back to the rationality discourse, Alba and Hutchinson (1987) argue that consumer’s inability to appreciate categorization will potentially result in suboptimal choices. Meaning that being responsive to the preferred categories will lead into a more optimal choice. This brings us to the second hypothesis:

H2: In a market that offers assortments of preferred goods, there is a positive relationship between the level of consumption and the product categories corresponding to each

consumer’s known preference

3.2. Segmentation

Thus far, this research has covered the effect of preference and level of consumption where the interaction covers both selective and general consumers. This section will dissect selective consumer independently. As the main concern of this research is to examine rationality as a function of preference by the means of examining selective consumer, a deeper analysis towards the attributes that make up selective consumers to begin with will be investigated further.

Furthermore, referring back to the consumer preference and assuming that selective consumer act accordingly to maximise their gratification and preserve their identity, selective consumer will consume more. Thus given the amount of selective consumer is significant in the total consumer composition, selective consumer consequently become the significant consumer in the market. Therefore, gaining insight on market segmentation within the selective consumers is advantageousTo gain more insights into the dynamic of selective consumer's consumption, this research will look into the role of segmentation and how does it affect the behaviour of selective consumer.

Market segmentation has been widely considered to be at the very heart of the consumer behaviour discourse (Morgan and Pritchard, 2001). Thompson and Schofield (2009) state that by directing the marketing efforts towards what is regarded as economically significant consumer, market segmentation has the ability to maximise demand. Thus, from management and marketing point of view, an exhaustive comprehension on how market segmentation affect demand and consumer's consumption is highly beneficial. As Kruger, Saayman & Ellis (2011) stated, not only limited to predicting consumer behaviour but understanding market segmentation is also useful for identifying market opportunity.

.

This research furthers the inquiry into consumer behaviour by trying to examine the different segmentations of the selective consumer and make use of the segmentations founded on some of the base variables summarised by Moscardo, Pearce and Morrison (2001), namely the behavioural and geographical segmentations.

The need to understand the pattern of consumption leads us to the first part of the hypotheses regarding segmentation where the behavioural segmentation within the selective consumer is examined through the proxy of repeat consumer – those who have previously interacted or consumed in similar or identical market setting. Some researchers have argued that behavioural variables make better bases for segmentation as they explain consumption behaviour better (Gladwell, 1990; Johar and Sirgy, 1995; Morrison, 1996). As Moscardo, Pearce and Morrison (2001) mention, in order to identify major markets, an a posteriori segmentation should be conducted. Therefore, an understanding from previous occurrence should be taken into account to better understand the effect of segmentation as type and frequency of consumption are the corresponding variables to examine behavioural segmentation.

Previous studies show that repeat customer possesses stronger relationship on product value compared to first timers (Lee, Lee & Yoon, 2009). Moreover, motivational values are considered to be paramount in the case of repeat customer (Paul et al., 2009) thus a positive effect is expected to be seen for repeat customer. This research seeks into whether selective consumer with previous purchase consume more in the current market. This generally means that market for selective consumer is segmented between those who have consumed the good before and those who have not. Indeed, if preference is consistent, selective consumer with previous consumption will likely to consume more as the consumption reaffirm their preference. Therefore:

H3: In a market that offers assortments of preferred goods, there is a positive relationship between repeat consumption and selective consumer level of consumption

Now, having taken into account the behavioural variables into analysing selective consumer behaviour, it is also important to keep in mind that there are also scholars that refute the behavioural segmentation approach in favour of the more conventional demographical or geographical variables (e.g. Davis, et al., 1993; Loker and Perdue, 1992). The main argument being, that geographical or demographical variables are statistically simpler to analyse, can be easily presented (Wilkie, 1994) and easier to use for practical application (Dodd and Bigotte, 1997).

One of the most prominent theory regarding geographical segmentation is the distance decay theory by McKercher and Lew (2003), where it is predicted that demand will rise to a peak at some perimeter relatively close to a market and descend exponentially as the distance increases thereafter. Indeed, understanding how the proximity of customer affects their consumption is important as it helps identify the obstacle that might hinder consumption

(Liang, Illum & Cole, 2008). Studies suggest that travel distance may vary depending on which activity someone participates in (Lentnek, Harwitz & Narula, 1981). In this regard, this research postulates that selective consumers are willing to travel the distance to reach the market. Nevertheless, beyond a certain point the number of consumer will inevitably decrease. This brings us to the second part of the hypotheses regarding market segmentation, where this research inclines to agree with the distance decay theory and predicts that the theory will hold for selective consumer’s consumption.

H4: Geographical distance to a market that offers choices of preferred goods has a curvilinear relationship with selective consumption

4. Empirical Setting

In this section, the empirical setting chosen to delineate the research question will be explained. This will start by examining the rationale behind the choice of creative industry and cultural good, specifically in the film industry and why it is a good fit for this research. Then the art film audience, subscription for unlimited cinema card in the Dutch market will be introduced as a measure of rationality and preference decision. Lastly, the setting of film festival is placed as the market for preferred good to further investigate consumer behaviour within an even more fixed setting that corresponds to consumer preference.

4.1. Creative Industry and Cultural Goods

The creative industry is often regarded as the extreme of industries, having puzzling labour markets (Menger, 1999), perpetual oversupply (Peltoniemi, 2015) and overriding heuristics in decision making process. It is fair to say that creative industry is the industry where passion and intuition took over rationality and economic expectation. In this sense, the agents of creative industries are willing to forego economic incentives – thus not maximising their utility – and embrace exceptional risk for the sake of their art.

The demand side of the creative industry has yet to be discussed thoroughly in academic literature. What little behavioural researches there are revolve more around consumer utilisation of media (Chuu, Chang, and Zaichowsky, 2009) or the growth of

creative spaces that create demands (Rogerson, 2006) and not consumer’s rationality or conduct. This gap in the demand side of creative industry discourse is very much alike that of the rationality and preference, making the two topics share a strategic commonality.

Ter Bogt, et al. (2010) reports that consumer preference of cultural good is relatively stable overtime. As mentioned earlier, when it comes to examining rationality as a function of preference there is a contention regarding the consistency of preference (Irwin and Payne, 1998). This attribute of consistency from the cultural good however, dispose of this problem and preference in the case of creative industry becomes an even variable for examining rationality.

In order to properly test the hypotheses in this research, there is a need to be able to distinguish different types of consumer properly while at the same time incorporating a good measure for examining consumption. Several studies have explored the taxonomy of different consumers in the creative industries and one of the most common indicator used is frequency of attendance (Cuadrado, 2000; Strang and Gutman, 1980, Peterson, 1980; Hodgson, 1992). Attendance is a good proxy for consumption in the creative industry because attendance is almost always tautological with purchase or consumption (e.g. attending a concert, films or exhibition). Indeed, returning to the identity discussion, Kleine, Kleine and Kernan (1993) mentions that how frequent activities are performed depends on the salience of the identity it represents. Consequently, the choice of creative industry as setting is very convenient as the preference and consumption is incorporated in one good measure; namely attendance.

Additionally, some studies have shown that consumer of cultural goods are not only preferential in their consumption, but also in location (Cuadrado and Frasquet, 1999). This coincide with the geographical segmentation behaviour that this research also tries to uncover. All in all, having all the desired features makes the creative industry the ideal setting

for the inquest of selective consumer behaviour and consumer rationality and preference in general.

4.2. Film Industry

In an industry where consumers display nonstandard beliefs and preferences as the creative industry, a deeper analysis of consumer behaviour is the stepping stone towards a better understanding (Eliaz et al., 2006). One way to achieve this is by exploring a specific audience within the industry. Gazley, Clark and Sinha (2011) and Watson (2006) note that academic studies often neglect film industry when discussing art and aesthetic consumption. Although as a product, films have been studied quite extensively, especially in regards to market categorisation (Krinder and Weinberg, 1998; Elberse and Eliashberg, 2003) and market performance (Gemser et al., 2007; Delmestri et al., 2005; Basuroy et al., 2003) an empirical research on audience is rare because of the scarcity of empirical data (Austin, 1984). All of which is very well since as Eliashberg (2000) states, predicting consumer demand in the film industry is tricky due to its experiential nature of the consumption.

Nevertheless, films are the quintessence of art and commerce coming hand in hand (Holbrook and Addis, 2008) and investigating drivers of consumer behaviour for film attendance provide assistance for better commercial performance (Eliashberg et al., 2006).

McKenzie (2012) highlights that different groups of film audience are interested in different cinema settings, making it even more suitable to the context of selective audience that this research takes up. The classification of audience that exist in the film industry is applicable for consumer rationality analysis as film audience are simpler in their preferential difference (Chuu, Chang and Zaichowsky, 2006) making them an even more fitting subject.

4.3. Art Film Audience as Selective Consumer

Selective audience in this research is defined as those who possess strong preference and consume accordingly. Subsequently, distinguishing different type of consumer is paramount to get meaningful results. In the film industry, contemporary researches have used different definitions to characterised the different type of audience on the basis of the film type and attributes with art and mainstream film being set apart based on the product characteristics such as budget (Geer, 1998), narrative structure of the films (Bordwell, Thompson and Ashton, 1997) and share of screens (Reinstein and Snyder, 2005).

This research will use the proxy of art film audience to examine the selective consumer behaviours. Most common descriptive definition of art film used in researches is the one by Irving (1995) which define art film as being produced by independent filmmaker, usually with limited budget and without star. Additionally, foreign films, documentaries and classic re-release are also generally considered to be art films (Twomney, 1956).

This research build on the division for art film as demonstrated by Gemser et al. (2007) whereby the dichotomy of art film and mainstream film is split on the basis of the cinema the films being released to. This division is based on film market identity (Zuckerman and Kim, 2003) which also corresponds to selective audience identity building attribute. To put it simply, in this research art film audience are film consumers who are committed to watch films in art-houses and smaller cinema.

4.4. Dutch Film Market and Unlimited Subscription Card

This is research examines the Dutch Cinema Market. Cinemas in the Netherlands are comprised of a mixture between chained multiplexes and art-house cinemas. Gemser et al. (2007) mentioned that audience of art-house films is much smaller compared to mainstream

films. This observation is consistent with the composition of cinema in the Netherlands, which in terms of capacity, share of revenue and share of visitors is dominated by big chained multiplexes (NVB, 2016; see appendix 1).

Most interestingly, in the Netherlands film market, there is an option for unlimited subscription card. Whilst big chained multiplexes have their subscription card for cinemas in their chain, art house cinemas are chained by an independent third party. Focusing on the art film audience, this research makes use of the commitment of unlimited subscription card for art-house cinemas in the Netherlands as a proxy to examines selective consumer rationality and behaviour.

In a way, subscription towards an unlimited access to art house cinemas also serves as an identifier that is independently chosen by the consumer. This corresponds to the identity building feature of preference and rationality, making the art film audience who subscribes to unlimited card become ever more suitable for testing the hypotheses in this research.

4.5. Film Festival

The last feature of the empirical setting is the market setting. This research will use the setting of film festival as it enables the behaviour of selective consumer to be put under a more constrained scope. Firstly, the setting of festival facilitates a line-up of products in a particular year (Paleo and Wijnberg, 2008), in this case the film festival features a line-up of art films. By definition the films feature in most festivals comes from independent production and therefore qualify as art film. This is particularly important because although art houses play art films, oftentimes they too feature non art film in their daily programme. Having a festival as a particular setting therefore makes the available option constricted into strictly art films and examination more thorough.

Furthermore, festivals are known to fulfil not only artistic endeavours but also social and economic roles. Archibald and Miller (2011) goes so far as saying that festival is itself a phenomenon, with its own economy. On top of that, ticket sales from audience are, in most festival, the most significant resources (Kruger, 2009), making festivals an ideal setting for examine economic consumption and consumer behaviour.

Particularly for film festivals, despite its exponential growth in the past decades, there has not been much published work that study film festival academically, this is quite regrettable as film festival works as a melting pot in assimilating different types of art films and has the value-adding benefit for art houses (de Valck, 2007).

In this study, the setting of film festival will serve as the market offer a variety of preferred cultural goods; namely art films. The aim is to examine whether in such setting where preference is facilitated there is a ubiquitous effect on being a selective consumer (art film audience) and level of consumption. This study will also investigate whether preference feature is in any way related to the product auxiliary such as categorization (by the proxy of art-film genre) and market segmentation (distance travelled to the festival and repeat audience).

5. Methodology

The research setting is a film festival in the Dutch film market and as mentioned, the selective audience is the art film audience who are subscribed to the unlimited card for art film houses throughout The Netherlands. The analyses and hypothesis testing will be execute using the archival data from the subscription card as well as the primary data from the ticket sales during the festival. Descriptive statistics will first give a general behaviour of the festival attendees and regression analyses will test if the number of visits corresponds to the theory and postulation in regards to rationality and preference tenets.

5.1. Card Subscription - Cineville

Cinevillepas is a subscription card which allows the holders to visit and watch screening (including festival screenings) at the corresponding art-houses and independent cinemas. These art-houses and independent cinemas generally have different program compared to the chained multiplexes operating in the Netherlands (e.g. Pathé, Vue, or Kinepolis) and emphasizes more on art films. Nonetheless, as previously elaborated, some of the films in the daily programs of these cinemas do sometime belong to the mainstream movies.

The monthly fee for a Cinevillepas is 19 euros and there are 37 art-houses and interim cinemas available across the Netherlands in which the subscribers garner unlimited access to every screening. The initial subscription for the card is for 4 months with automatic renewal unless the holder request cancellation. Upon ordering the card, holders will be required to pay the first month of subscription on location. The subsequent monthly payments are deducted directly from the holder’s bank account. When desired, cancellation can be done through email and is free of charge after the required subscription period.

The normal full price ticket for a single screening in the corresponding art houses is between 4-12 euros with an average price of 9.15 euros per screening (see appendix 2). These prices exclude discounts for students and other pricing options. The data for individual cinemas are collected manually using the available website for each art-houses. Now, returning to the basic utility maximization theory, consumer would be rational to subscribe if they watch at least 2.07 films per month, otherwise they would be worst off. However, building up from the mere utilization theory; this research postulate that the feature of preference – that of towards art film and attending screenings at art houses – will amplify the attendance of the cardholders.

Additionally, the setting of film festival will isolate the subscriber into a more constrained setting and therefore their behaviour as selective consumer can be further investigated. The data acquired from Cineville includes of detailed attendance (time, movies, which art house, etc.) as well as control variables (such as age, gender and location). The data that is released for the purpose of this study consists of all subscribers (over 20,000 in the year 2015) and exclusively for the month of March in 2016 where the festival, which will be explained further in the following section, took place. The identities of the subscribers are kept anonymous with only card number released to identify each individual.

5.2. Festival - Cinemasia

Cinemasia is a pan Asian film festival which takes place annually in Amsterdam. Every year, the festival screens various full length feature, short films and documentaries focusing on Asian diaspora and culture over the span of six days. Tautologically, the fact that the films screened during Cinemasia are foreign (Asian) and/or independent films, this festival fit to serve as a market which provide abundant preferred goods; namely art films.

In addition to the main festival, Cinemasia also organized quarterly screening in different cities in the Netherlands. To date, the festival has organized nine film festivals with over 4000 visitors in 2016.

This research makes use of the data of the festival in the year 2016 where the festival took place in Kriterion and Rialto Amsterdam, both of which are art-houses included in Cinevillepas subscription. Therefore, the Cinevillepas holders are able to watch screenings at Cinemasia without having to pay additional charges. However, the data for this research focus on the screenings at Kriterion. Moreover, it is important to mention that there are different type audiences that attend the festival and this research only examine paying audience. There are significant number of the festival visitors who are business partners of

the festival as well as those who receive special invitation as guests thus gaining free entrance to the movie screenings throughout the festival, these types of audience are not being included in the analyses.

5.3. Sample and Procedure

The quantitative research is executed using consolidated archival data of Cinevillepas database and film attendance during Cinemasia Film Festival. The data from Cineville consist of detailed ticket purchased for all subscribers’ visit as well as the general information of each individual card holder. The data from Cinemasia is consists of detailed ticket sales with from the box office at Kriterion Amsterdam and ticket purchased via the festival online website for screenings at Kriterion.

Initially the data from Cineville consists of 59,249 observations for the month of March 2016, these are then filtered with the location of Kriterion during the six days of the festival and by crosschecking the films and make sure only the festival films are included in the observations. This yield 778 visits from 451 different Cinevillepas holders. Then combined with the box office and website sales data for paying audience yields 1459 visits from 1012 different individuals. The first and second hypothesis make use of the this fully combined data for Kriterion. Each individual in this dataset possess a unique identifier which comes from either the Cinevillepas number or from the transaction code from the website and Kriterion box office. The number of attendance for different screening is then counted and the number of screenings is used as the dependent variable for this research.

To examine the product categorization aspects of this research, the films screened are categorized in terms of their genre. Sedgwick (2002) mentions that although its fluidity makes genre usage as a working categorization unattested, its usage serves the purpose satisfactorily. For this research, from the total 38 films being screened in the festival, the

categorization is divided into three main genres for full length film in accordance to the first genre mentioned in the festival guidebook; namely: drama, comedy or musical, and documentaries. Then, the number of attendance is counted for each different screening per film genre that an individual attended, categorizing the audience consumption to the product category it belongs to.

Ever since the dawn of the film industry, films as a product has attract audiences differently; with general audience more interested in mainstream such as musicals or films with high spectacle (Docherty, 1995). Following this perspective, categorization serves as additional marker of preference since art film audiences are usually associated with particular type of genre like documentaries, foreign movies and serious drama (Smythe, Lusk and Lewis., 1953 as cited in Chuu, Chang and Zaichowsky, 2009). On top of that, Chuu, Chang and Zaichowsky (2009) also mentioned that art film audience form an identity to their consumption behaviour, this support the identity framework of selective audience.

In addition to the Cinevillepas, the festival screening also offers concession for student, elderly people, city pass holder and youth card holder. However, whilst Cinevillepas holders are granted free entry, the other concessions receive discounts between 0.5 to 1 euro per screening making the effect negligible and will only be featured as supplementary attributes for analysing behaviour, along with variables to indicate the time slot and screening times.

Whilst the analyses for the first and second hypotheses make use of the complete data (dataset I) for all paying attendees, for testing the third and fourth hypotheses, only the data from Cinevillepas holder is used (dataset II). This data is again isolated to the time and place of the festival. However, for this set, there are more individual details available. These details are not retainable in the consolidated data as box office sales cannot provide background data for each consumer/attendees. The data of different individual cardholders during the festival

are then cross checked with the data from the attendance of previous edition of the festival as well as other screening by Cinemasia during the year 2015. This is possible by the means of the subscription number identifier and because other screenings by Cinemasia have also taken place in Cineville corresponding art houses. There are 505,730 observations in the year 2015, after being filtered out 102 cardholders are identified as having attended the festival and/or special screenings by Cinemasia before. From 451 individuals in the second dataset, initial cleaning to exclude errors and sort out any missing elements left 443 individuals eligible for the analysis.

5.4. Variables and Measures

The variables used in this research is generated from all the archival data granted by both Cineville and Cinemasia. After the decoding process which include counting and transforming raw data into binary and ordinal data, STATA 14 is used for all statistical analyses. There are 13 variables used particularly to test the first two hypotheses (for detailed description of every single variable check appendix 3).

Consumption. The total numbers of screenings and screening per genre attended during the festival is used as the dependent variable. For the first hypothesis the total number of screenings of all audience is used. The second hypothesis makes use of the total number of screenings per genre for all audience. For the third and the fourth hypotheses, the total number of screenings for all selective consumer is used (dataset II). These variables are count variables, meaning it only takes discreet integer values. The use of ordinary least square regression for the analyses is therefore not desirable and potentially problematic (Coxe et al., 2009), considering the outcome is generally low with mean of 1.441 in the first dataset and 1.723 in the second. Thus, Poisson regression is used instead.

Selective consumer. Cinevilepas possession serves as the proxy to identify the selective consumer from the rest of the visitors of the festival which are subsequently divided into those who pay the full price and those who receive concession price. In addition to that, another two variables are added for the extension of the model to indicate the time of the screening, be it matinee or primetime and screening in the weekend and weekdays; all of which is indicated by the generation of dummy variables.

Product categorisation. The variables for product categorization are divided into the genres (drama, comedy or musical, and documentary) with each variable indicates the number of films watched by each individual that is categorized as either one of the four.

Behavioural segmentation. The behavioural segmentation is analysed using the independent variable repeat visitor which indicate whether the consumer has previously attended Cinemasia festival screening before or not. The festival screening meant in this case is previous festival in the year 2015 and quarterly screenings which are also organised by Cinemasia.

Geographical segmentation. To analyse the geographical segmentation, the use of consumer's/ card holder's postcode is used. The interval for the geographical distance is measured by the proximity of consumer's location adjacent to the festival location at Kriterion Amsterdam (for more detail check appendix 4). At this moment, with skewness of 0.3776, there is no evidence to view the variable postcode which measure for geographical distance to be other than normally distributed. However, the kurtosis of 3.491 suggest that the distribution is rather heavily tailed which is indicative of the relationship suggested in the fourth hypothesis.

Control variables. For dataset II, two constants are available to serve as control variables; namely age (the age of the cardholder at the time of the festival) and gender (binary – male/female).

Lastly, multicollinearity assessment is performed by computing the extent to which the standard error of the coefficient of interest variance has been inflated. This is done through examining the variance inflation factor (VIF). For this assessment the log variables for the dependent variables are generated to enable linear assessment. For database I and II case with VIF mean of 2.01 and 1.02, multicollinearity is not a problem.

6. Results

This section presents the result for analyses and hypotheses testing. First, the descriptive statistics from both datasets will be described. Second, the correlation analyses are presented. Lastly, the result from the regression analyses is explored.

6.1. Descriptive Statistics

In this section statistics about variables are provided. Only the relevant findings are described below, for more detailed description refer to appendix 5 for the descriptive summary of the sample in the first dataset used to analyse the first and second hypotheses and appendix 6 for descriptive summary for the sample in dataset II which is used to analyse the third and fourth hypothesis.

Most consumer watch one film during the festival (77.87%) and unlimited card subscribers comprises 44.57% of paying audience in Kriterion with average consumption of 1.724 screenings per card holder. Considering the monthly fee which require 2.07 films to be consumed per month to breakeven with the cost of consumption, and that consumption is discrete; card holder consumption's during the six-day span of the festival is favourably optimal. More than half of the screenings attended by consumer take place during primetime (after 16.30 - 62.94%) and 57.81% of the films are watched during the weekend.

For the second database, although the percentage is somewhat lower, most of the consumers still watch one screening (69.07%). The majority of consumer lives in inner city of Amsterdam (56.21%, 85.78% if combined with the same and neighbouring postcode are of Kriterion) and 21.9% of the consumer has visited the festival or other screening by Cinemasia beforehand.

6.2. Correlation

Table 1 shows the correlation matrix table with mean and standard deviation for every variable in dataset I using the one-tailed Pearson's correlation, and table 2 shows for dataset II respectively.

The mean for total number screening watched per individual is 1.44 for dataset I and slightly higher in dataset II at 1.70 (with minimum 1 screening and maximum 12) with standard deviation of 1.14 and 1.46. Table 5 shows that there is a positive and significant correlation (r = 0.222 and p<0.01) between having the Cinevilepas and the number of screening watched during the festival. As for other concession price, all show somewhat negligible effect with only the elderly discount shows a significant correlation with the number of screenings watched with positive r of 0.088 and p<0.01, while student discount shows negative effect at r of -0.071.

The correlation matrix between independent variables from dataset I shows some interesting findings, for example the possession of Cinevillepas is shown to be positively correlated to the number of screenings attended during matinee (r = 0.279, p<0.01). A significant and positive correlation can also be seen between the number of documentary films and daytime screening with r of 0.453 at 99% significance level. Moreover, the number of screenings during daytime shows significant and negative correlation with paying full price (r = -0.165, p<0.01).

Table 2 shows positive but small and non-significant correlation between having visited the festival before and the number of screenings watched (r = 0.044). The geographical distance and the number of films watched is negatively correlated for both normal and squared value (r = -0.038 and r = -0.013, resp.), having the squared value for this variable. Finally control variable age and gender shows small but positive correlation with the number of screenings attended during the festival (resp. r = 0.098 and r = 0.068); with age having a significant effect at 95% significance level.

Table 1 – Correlation Table Dataset I

Table 2 – Correlation Table Dataset II

6.3. Hypotheses Testing

Table 3 presents the result for the regression analyses. In the first model used to test the first hypothesis, the variables of card ownership cineville, and attendance timing

noweekend and nomatinee are estimated. Additionally, other type of audience is also added

into the model; namely: the audience who pays fullprice tickets (fullprice) and other concession (student, cjp – youth pass, stadspas – city pass and elderly discount – 65+). The McFadden Pseudo R-squared of 0.1365 show that the model has a reasonable fit, although not particularly very strong. The coefficient of each variable, except for the youth discount however; show a significant relationship at z<0.05;0.01. Indeed, model 1 shows that having a subscription card has a positive and significant effect towards the number of screenings watched. Moreover, compared to the rest of relationship with the type of tickets the consumer buys (bearing in mind that this assessment being analysed using dummy variables), being art film audience marks a significantly higher relationship with the number of screenings watched. Therefore, with possession of the card (cineville = 1), the difference in logs of screening attended is expected to change by 75.91%. This analysis show that hypothesis 1 can be supported with Poisson regression that yielded statistically relevant positive relationship between being a selective consumer and number of consumption.

Table 4 presents the result for the regression analyses for examining the second hypothesis. In these regressions, the dependent variables are the number of screenings attended for each genre per individuals. The point of reference for the analysis in hypothesis 2 departs from the earlier remark by Smythe (1956) and Twomney (1956) which indicate that in terms of product categorization; art film audience leans toward drama and documentary genre.

There is indeed a positive relationship between card possession (marker of being an art film audience and subsequently selective consumer) and the number of screenings per genre. This is expected quite considering that as hypothesis 1 illustrates above that there is a significant positive relationship between being a card subscriber and the amount of the screenings attended during the festival. Therefore, the positive and significant relationship at the model for drama, comedy/musical and documentary in Table 4 is consistent with the previous finding.

The amplitude of the coefficient cineville for each different genre, however, suggest that for genre drama being a selective consumer do not necessarily amplify consumption for the categories corresponding to art film audience's preference. Furthermore, the regression also shows a negative relationship between student audience with documentary genre, and youth audience with comedy/musical genre. The former is consistent with the age demography of art film audience suggested by Cuadrado and Frasquet (1999) where art film audience tend to come from the older demography; however, the latter proves to be quite contradictive. Although the relationship in this cases are not significant and as previously stated the other concession types are quite negligible, this finding provides an interesting perspective on consumer's preference consistency and a deviation from the prediction. Therefore, on the basis of this regression analysis, the second hypothesis cannot be supported

as the positive relationship between number of consumption for each specific genre corresponding to audience's know preference and audience type is not entirely consistent with the predictions drawn from the preferences.

Table 4 – Poisson Regression Result Dataset I - Model 2,3,4

For dataset II, the first step a control model with gender and age variable is estimated. The model shows that age has a small but significant effect on the number of screening watched during the festival for subscription card holders (z<0.05). The next model tests the relationship between behavioural segmentation by the proxy of repeat visitation to the festival and the number of screenings watched by the subscription card holders. The regression results from table 5 show a small and insignificant result for the dummy variable

repeat which signifies whether the consumer is a repeat consumer or not. Since the Poisson

regression cannot provide a sufficient support that the relationship between having visited before and the number of screening is significant, hypothesis 3 is thus rejected.

Table 5 – Poisson Regression Dataset II

In order to test the curvilinear relationship between the geographical distance and the number of screenings watched by Cinevillepas holder, both the number of total screening on the basis of consumer's location and the average number of screening per individual per area are examined.

Graph 1 shows the distribution of selective consumer's geographical distance in relation to the number of screenings visited throughout the festival. The number of total screening peaked at area 3; meaning that in aggregate, consumer who lives in inner Amsterdam area consume the most during the festival. Indeed, in accordance to the decay theory by McKercher and Lew (2003), the number of screening increased by distance from area 1 to area 3 then decreased significantly at area 4 to over ninety percent decrease, though the number somewhat increases for consumer who lives in area 5.

Graph 1 – Total Number of Screening per Area

Graph 2 illustrates the average screening watched per individual per area, it shows that in terms of intensity, individuals coming from area 4 consume the most during the festival with the highest average of 2.02 whilst individual subscriber from area 2 consume the less. Thus, there is sufficient evidence that support the decay theory and the curvilinear relationship between geographical distance and consumer consumption.

Graph 2 – Average Screening per Individual per Area

77 186 394 38 82

Area1 Area 2 Area 3 Area 4 Area 5

1.9175

1.56

2 2.03

1.82

7. Discussion

This study began with the discussion about rationality and consumer preference and consumption. Whilst economic theory of consumer demand often neglects taste and preference (Potts et al., 2008), consumption assessment based on mere utility maximization theory proves that consumer behave irrationally (Della Vigna and Malmandier, 2006; Cohen and Dickens (2002). Previous studies have yet come into a consensus on how best to bridge this discrepancy, especially with most of the studies fail to examine the demand side perspective (Markusen et al., 2008). Based on the aforementioned discourse, this study takes upon the demand side and examine consumer's rationality as a function of their preference.

The basic idea of this research is that consumers who embody fully their preference, be it for the reason of self-identification from consumption or otherwise, will consume rationally. The concept of selective consumer is then introduced whereby consumer possess strong preference towards certain goods corresponding to their preference (dubbed as preferred goods). In order to examine the behaviour of selective consumer a specific setting is then introduced so that the market available in this study offer abundant assortments of preferred goods. The aim of this setting is to amplify the effect of strong preference on consumption and in order to be able to draw better distinction among consumer. Then the setting of film industry with art film audience and film festival is introduced as a proxy for selective consumer and the market setting.

Departing from the literature gap and further discourse, it is expected that being selective consumer will have positive relationship with the amount of preferred good consumed in a market where assortments of those goods are abundantly available. This proposition is supported by the hypothesis testing from festival screenings which presents a

significant result with high distinction between selective consumer and other type of consumers in number of experience goods consumed (films watched).

Moreover, the auxiliary of consumer consumption is examined to better grasp consumer behaviour. The second hypothesis expand on the preference feature and postulate that given clear product categorization in the market setting, preference will be consistent with the preferred categories associated with different types of consumer. In this case the proxy of film genre is used and art film audience is associated with intense drama and documentation genre. There is no definite evidence to support the postulation that selective consumer will consume more of the preferred goods corresponding to their known preference. Moreover, demographic differentiation among consumers also suggest deviation from the rigid genre categorization. One explanation towards this respect is that the demographic segmentation of art film audience is in itself quite contestable (Möller, 1983), therefore audience member's sensitivity towards genre categorisation might be slightly

compromised.

The third and the fourth hypothesis deal with the feature of segmentation within the selective audience. The behavioural segmentation examined using the proxy of repeat consumption angle is proven to be unsupported as there is no significant evidence from the regression result to support that selective audience's previous engagement in the market holds effect on their current consumption. One reason to better reconcile with this result is by understanding that consumer preferences evolve (Simonson, 2008) overtime and that audience composition is volatile and likely to change over the course of time (Dempster,

2006).

The fourth hypothesis deals with the geographical segmentation within the selective consumer. In terms of number of consumption, the distance decay theory by McKercher and Lew (2003) proves to be consistent with the findings with a constant rise up to a certain point