Diffusion of artists in Corporate Art Collections : an exploratory study on herding behavior in small-world networks

(1)

Diffusion of artists in Corporate Art Collections

(2)

Statement of originality

This document is written by Eliaan Elisabeth Blijdenstein who declares to take full responsibility for the contents 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.

(3)

Acknowledgements

“The important thing is not to stop questioning. Curiosity has its own reason for existing. Anyone who has never made a mistake has never tried anything

new”. – Albert Einstein

“Replace fear of the unknown with curiosity”. – Penelope Ward “I don’t know where I am going, but I am on my way”. – Carl Sandburg “We are always in the process of approaching truth and perfection without

actually ever reaching truth or perfection”. –Mark Manson

I would especially like to thank Monika Kackovic, the Assistant Professor in Entrepreneurship and Innovation and Postdoctoral Research and Fellow in Entrepreneurship and Innovation of the University of Amsterdam. I don’t know how to thank her for guiding me into the right direction every time. With

her patience, guidance and comprehension I was able to complete a thesis that I enjoyed writing. I would also like to thank her for correcting my drafts. I would particularly like to thank Alan Bloor (Marketing Expert), Hayley Marx

(Artist), Alexander Gullon (Digital Consultant) and Mino Blijdenstein (BPO Consultant) for reading and correcting my thesis as well as giving me insights and

new perspectives.

The members of my family and friends have played an important role in the fulfillment and support of this thesis. I would like to thank my parents, whose love

and guidance are with me in whatever I pursue. This work would not have been possible without the financial and emotional support of my family, especially my mother and father, Johanna and Mino Blijdenstein. They are my ultimate role models.

! ! ! ! !

(4)

1. Abstract

! Considering the case of diffusion of artworks across corporate art collections (CACs) this paper investigates firm characteristics on herding behavior of curators in the primary art market. This research constitutes a unique dataset that includes the purchases of artworks of a closed association of CACs in The Netherlands. The specific characteristics used in this research are an (1)-industry component and (2)- experience in collecting art. This research controlled for firm size and standardized legal form of the firm. A Poisson regression and a Multinomial Logit Regression (MLR) model were used to understand the patterns of the data collected. Drawn from simple statistics, there is empirical evidence that herding occurs within the sample studied. There is partial empirical evidence that herding behavior occurs within some of the industries researched in this study. Only in the health and financial industry. It was found that the Public Sector firms did not herd as much as the commercial sectors. Drawing from the results it cannot be inferred that firms with most experience collecting art act most independently, as to buying decisions of artworks are concerned, compared to those firms with less experience in collecting art. This finding suggests that there are cliques of experts that do herd with each other. This paper ends with a call to more empirical research on the subject of herding behavior in corporate art collections. Further research is suggested to be able to generalize these findings.

Key words: Corporate Art Collections, Primary Art Market, Herding, Small-world Networks, and Recurring.

(6)

2. Introduction

This study investigates herd behavior in the context of corporate art collections in The Netherlands. A Corporate Art Collection is defined in the literature as the compilation of any type of art held by companies as a collection (Kottasz, et.al 2007). The aim of this study is to gain a better understanding of social network structures that can be characterized as a small world network (Watts, 1990; Burt, 2004; Sullivan, Tang and Marquis, 2014; Uzzi and Spiro, 2005), defined as a network with high clustering and short average geodesics, a combination that is characteristic for many real social networks (Schnettler, 2009). Particularly how these types of networks can trigger this herd behavior (Raafat et.al, 009; Bikhchandani and Sharma, 2000; Bikhchandani, Hirshleifer, and Welch, 1998). This paper addresses the research question of what firm characteristics have an influence on the observed herding behavior in small world networks. This paper looks into the dynamic patterns of curators buying behavior, leading to the phenomenon known as “herding”. Raafat et.al (2009, p.420) describes herding as “a form of convergent social behavior that can be broadly defined as the alignment of the thoughts or behaviors of individuals in a group through local interactions without centralized coordination”. The group in subject in this paper is the group of curators belonging to the Vereneging Bedrijfcollecties Nederland (VBCN). In this paper, the focus is on herding behavior in Corporate Art Collections (CACs) in the Netherlands. A better understanding of “herding behavior” in the CACs is needed to be able to get more insights about the primary art market. Understand different patterns of curators behavior is also important for managers of corporate art collections. Curators are able to assess the quality of an artwork and trending artists through the behavior of others. This paper

(7)

has the objective to assess which of the patterns can be witnessed in those firms that behave similarly. Does a particular type of firm signal a higher chance of imitating another? This will allow curators to be more conscious about their buying decisions in the future. This is the first study that identifies herding patterns in Corporate Art Collections. This paper looks at two firm characteristics, firm industry and experience in collecting art. This paper is of importance because herding has been a widespread term in the financial sector, however little attention has been paid to herding behavior in small-world networks and more specifically in the cultural industries.

Herd behavior is also defined as an investor’s tendency to ignore their own judgment and to mimic the decisions of other people (Bikhchandani and Sharma, 2000; Angela-Maria, Angela-Maria, and Miruna, 2015). Herding behavior is most commonly used and understood in economic and financial settings (Fracassi, 2016; Hirshleifer and Teoh; 2003; Raafat et.al, 2009). Especially after the financial bubble that burst in 2007 and the bank runs scandal. This term has often been studied in settings such as financial stock markets (Fracassi, 2016; Bikhchandani, and Sharma, 2000; Angela-Maria, Maria, and Miruna, 2015) and more recently in the creative sector. For example in music ratings (Salganik and Watts 2008) and movie reviews (Lee, Hosanagar and Tan 2015). Further evidence of herding can be found in studies about online product recommendations (Senecal and Nantel, 2004), and in online product popularity (Ward and Putler, 1996). Raafat et.al (2009) reviews the extensive range of theoretical frameworks used in past literature on herding behavior. It has been present in many different disciplines ranging from economics to sociology and political science. Georg Simmel (1950) was the first researcher that considered social networks when studying herding in the discipline of sociology. This researcher describes how societies at

(8)

different times (i.e. middle ages vs pre-modern societies) have different structures of social ties. The argument Simmel uses is that membership of a group provides benefits. Similarly to studying corporate art collections and herding behavior it can be noted that imitating other curators can cause benefits to the imitators. These benefits include quality certainty and risk-averse purchases of paintings for a firm’s collection. Granovetter (1973) explored the strength of weak ties, which are defined as interpersonal ties that carry information through connections between people (Raafat et.al, 2009). Social network in diffusion processes, such as the diffusion of artworks, stressed the importance of weak ties, which permit this diffusion to actually take place (Raafat et.al, 2009). Weak ties allow individuals to reach people that are further away from them in the network. In the VBCN association, all the members are tied to each other through this association, which allows for any individual that is part of the network to reach another one, either through other individuals belonging to the network or the board of the association. Allowing them to diffuse information about what acquisitions have been made during the existence of the VBCN, since the year 2005, its foundation year. Therefore this network setting and situation allows a coordinated action to buy the same art. Since it allows information to flow between nodes belonging to the network simultaneously. Benerjee (1992) sets up a model in which paying attention to what everyone else is doing is rational because their decisions may reflect information that they have an others do not. This leads to a likely consequence of people trying to use other people’s information, which eventually causes herd behavior - everyone is doing what everyone else is doing- even when their private knowledge suggests doing something different (Banerjee, 1992). One of the most recent studies concerning social network analysis and social behavior

(9)

convergence is a longitudinal study of 20 years done by Fowler and Christakis (2009). This study shows that heppyness spreads through social networks, peoples actions and behaviours are contagious. A social network is a social structure made of nodes – people- that are connected by one or more specific types of interdependency (Raafat et.al, 2009). The interdependency in subject in this particular research is the fact that all these individuals collect art for the purpose of an organization. Understanding the pattern of interactions, which is connected to whom, can clarify the role of behavioural strategies. Even though in this paper the research does not go as far as exploring the complete links between all the curators, interviews with curators have been done, that lead this study to believe that there exists a pattern of interactions in the VBCN.

However, little research has been conducted on this type of behavior in a specific type of network; small-world networks. Within the small-world research there are not many studies that empirically trace diffusion of information processes (Schnettler, 2009). The characteristics of a small world network are locally clustered into dense sub-networks or cliques that are sparsely connected, linking network members through a relatively small number of intermediaries (Baum, Shipilov and Rowley, 2003). Most collective behaviours spread through social contact; social networks are the pathways along which social contagions propagate (Centola and Macy, 2007). Studies of diffusion dynamics have shown that structure of a social network can have important consequences for the patterns of collective behavior (Newman, Barabasi and Watts, 2006). Therefore it is interesting to study herding behavior in these types of networks. Small-world networks can be a way to seek an explanation for coordinated behavior. Coordination of collective action has been studied in some

(10)

fields; however not so much in relation to the corporate art world. Schnettler (2009) provides an overview of 50 years of small-world research. The current state on the literature of small-world networks includes literature on three dimensions; (1) structural, (2) related and (3) psychological. Which further splits process-related in two parts (2a) diffusion and (2b) search. The study of herding behavior in corporate art collectinos focuses on the process-related dimension of netoworks in which social processes occur through channels of contacts (Schnettler, 2009). Diffussion and search are two social processes that are of importance in the study of corporate art collections and artists. Pool and Kochem (1978) considered the pioneers of the small-world concept offer a few examples of social processes such as the diffusion of political ideologies and the stratetigc attempt of individuals to gain access to influencial social circles to obtain important information through their network channels (Schnetttler, 2009). The small-world network literature arised from the typical encounter of two strangers having common acquaintances. This part is important because it is the foundation of why the diffusion of artworks is supported in this paper. Curator’s talk to one another, casual encounters where new acquisitions are discussed, word of mouth, etc. The network this paper studies, by its small-network characteristics, allows information gets to others very rapidly and leads to a convergent action. Easily difussing the information through the network by short path lengths. Later on Pool and Kochem started to create mathematical models to explore the connections between individuals. Another contribution was the Burt and Uchiyama who examined the relative importance of network communication versus network position in regard to the diffusion of medical innovations (Schnettler, 2009). This finding of Burt and Uchiyama is interesting because it supports the idea that

(11)

information of innovations diffuses rapidly between firms in an industry. In the same way this paper uses new artist information that rapidly difusses. Doreian (1984) suggested a method to implement network data and individual characteristics in network autocorrelation regression models (Schnettler, 2009). Regression models therefore will be a core aspect of testing the hypothesis and patterns in the data in this paper. There is a distinction made in the literature regarding small-world effect and

small-world network. Newman describes the “small-world effect” as “the fact that

most pairs of vertices in most networks seem to be connected by a short path through the network” (Newman, 2003; p. 181). A small world networks is a term defined by Watts and Strogatz (1998) as a class of networks that share high clustering and short average path lengths. A path length is the number of paths that connect nodes which one individual needs to go through in order to reach a particular node. This paper uses this underlying theory at the group of curators level to explain the effect of the diffusion of artoworks. The respective characteristics of small networks have been studied in networks such as scientific collaboration (Newman, 2001), ownership links among german firms (Kogut and Walter, 2001), and actors co-appearing in several movies (Uzzi and Spiro, 2005). The current literature has not yet found an approach to explore the dynamics n the primary art market and corporate art collection networks. The VBCN is a small and scalable network with clear boundaries, which allows the exploration of collective purchases patterns present in the several collections. Many real social networks have been found to be scale-free and to display short chains characteristic for the small-world effect (Schnettler, 2009).

Finally, the primary art market conditions are also a crucial factor that reinforces the herding behavior observed. The primary art market is rife with uncertainty about

(12)

quality because of the oversupply of artists, no clear objective evaluation about quality, and uncertainty in demand (Caves, 2000; Menger, 1999). Curators are assumed to be expert buyers that are risk averse, so in addition to searching for information through their network, they also engage in observational learning or herding behavior. Which is a behavior that will help resolve any accountability issues they may have in the organization as well as to decrease the uncertainty that is comprised in quality in the primary art market. The more CACs buy an artwork of an artist, the more reduction uncertainty there will be about this particular artist regarding quality, given the signals that others are giving in engaging to buy this artist's artwork. Creating a self-reinforcing feedback loop.

This study has collected a unique dataset composed of quantitative as well as qualitative data. The quantitative data was gathered from the VBCN (2017) website and the firms websites that are members of the VBCN. The information gathered mainly comprised the artists in these collections. On the firms website further information was gathered about the artists featured in their collections. This paper focuses on this data and in studying the patterns of purchasing behaviors, specifically organizations that buy from the same artist. This dataset is very relevant because it is assumed that those artists featured in these websites are the ones the organizations are most proud of. This information allowed this study to collect 3053 cases of artists present in 49 firms. Additionally to this dataset, 4 qualitative interviews where done with 4 curators of the VBCN to gain further insights into the links between nodes in the network, common practices, common grounds, galleries they usually go to, etc. As well as budget information and places they go to buy art.

(13)

First a simple Poisson regression model will be used to understand the overall impact of each independent variable. Then a multinomial logit regression (MLR) will be used in order to investigate the ability of the independent variables to predict the probability of groups of firms having recurring artists with all the variables held constant. MLR modeling is, in fact, used for data in which the dependent variable is unordered or polytomous (i.e. values are more than two categories) and the independent variables are continuous or categorical predictors (Badri, Toure, and Lamontagne, 2015). The dependent variable in this paper, the recurring artists are grouped in recurring none, twice and up to nine times, representing 9 different categories of artists. The recurring artists variable has been compared to the two important independent variables: experience measured in years and the type of industry they are in. The experience variable addresses which curator’s leads to a set of coordinated actions. The industry variable addresses the diffusion to nearby neighbours, to try and understand if close competitors act alike. This study predicted that the more years of experience the less herding behavior would be observed. Firms in the same industry were also predicted to herd with each other more than with distant industries. Surprisingly the outcome we predicted about experience was quite in the opposite direction. Those that are most experienced herd with eachother. Industries in some of the sectors herd with each other and others do not, this hypothesis was partially supported.

In this paper we address a gap in the literature concerning herding behavior in light of small-world networks and in the setting of the primary art market. This paper adds to the current state of literature by studying on an understudied empirical domain. Herding behavior in the creative industries. The core aim of this paper is an

(14)

exploratory research to gain finer insights and grained understanding of the patterns found in the buying behavior of curators in The Netherlands belonging to the VBCN. This paper adds to the literature of human herding behavior patterns in a specific type of social network. Adding to the current research of pattern-based approach to studying herding behavior defined by Raafat et.al (2009). Which is by the ‘reductionist’ approach. First making some assumptions in relation to why this phenomon takes place, and after, observing the patterns of those that express this particular behavior. In this case curators have been modeled as simple imitators. This means that these particular individuals have been treated as well-defined and simple individuals that yield herding behavior. So the focus is on patterns and not on people (Raafat et.al, 2009). Even though it is an abstract approach to herding it allows researchers to better understand human behavior.

This paper will be organized as follows. First of all the research question will be stated together with the general framework of this study. What patterns are observed in the firms that have recurring artists? What factors do positively or negatively influence herding behavior in small networks? Then this paper will start with a literature review of CACs, herding behavior, small world Networks and the primary art market conditions from which the main assumptions of herding behavior have been drawn. After this, the hypotheses will be introduced. Once the hypotheses have been backed up with relevant literature, the method will be explained. A multinomial regression analysis will be made with one dependent variable that represents herding behavior consisting on the count of recurring artists. Artists with higher recurring values are regarded as more popular through the collections. The independent variable will be the years the firm has been collecting, the industry that the firms that manifest

(15)

this behavior are embedded in. There will also be two control variables; a size indication based on the number of employees that work for each organization and the legalized standard form of the organization – profit or non-profit. An analysis of the results will be thoroughly examined and the hypothesis results will be drawn from these results, comnbining the outcome of the interviews. A discussion will follow reflecting on the theoretical and managerical implications of this study. Finally the Conclusion of this study will be drawn and the limitation of the study will be overweighed.

2.1 Research Question

!

What types of firm’s characteristics have an influence on herding behavior in small-world networks?

In Figure 1 a visualization of the general model this paper is going to follow. This framework will help firms that collect art be more aware of the buying behaviors and the patterns that are observed of the group of curators in subject. What type of firms are more likely to herd, compared to those firms that are less likely to herd. The model proposed will be elaborated in the coming chapters. Basically this model considers simple factors such as industry and years of experience, which allows this study to predict the probability, which type of firms tend to herd more with each other compared to those that choose to have more unique pieces of art in their collections.

(16)

Figure 1 – General Model

3. Literature review and hypotheses

It is difficult to directly observe the decision process of curators. Decisions about buying an artwork over and other can be due to decisions that have been made more unconsciously and others more consciously. One way to work around this issue is to focus on the implicit herding incentives of the curators in the CACs context. This chapter’s purpose is an attempt to explain the patterns of herding behavior in the context of corporate art. First of all the literature about corporate art collections will be shortly reviewed. Then three levels will be used to discuss the assumptions of this study. The first level is at the firm level, at the level of the curator, the second level is the association and the network of curatrors, and the third level is how these two

Proposition H1: Curators are held accountable for the purpose of the collections of the firm.

Proposition H2: The VBCN shows characteristics of a small-world network

H3: Herding (Count -measure recurring)

Poisson Regression Multinomial Regression

Industry Industry Experience _Collecting

H4 (+) H5 (-) H5 (-) Experience Collecting H4 (+) H3: Herding (Count-measure of recurring split in categories)

(17)

assumptions can be leading factors of herding. Finally this section focuses on the primary art market level.

First the literature about herding behavior is examined and how it can be applied to curators and corporate art collections. The next section will look into the VBCN network qualities through the theory of small-world networks, and how this setting can cause herding behavior. The last section is an attempt to explain the behavior of curators buying decisions observed in the recurring artists in different collections through the market conditions of the primary art market. The purpose of this section is to try and understand, through the use of literature, the observed behavior by making implicit herding incentives assumptions. In order to later on explore possible patterns of firms that manifests the same buying behavior.

3.1. Corporate Art Collections

Art collecting as a corporate endeavor was introduced by Thomas J.Watson, the founder of IBM, in the early 1930’s (Behnke, 2007). Short after in 1956 David Rockefeller, considered the pioneer of modern corporate art started his art acquisition program in JPMorgan (Sharf, 2012; Sheldric and Dhir, n.d;). Where contemporary art started to be acquired by firms. Nonetheless it was Thomas J. Watson the first one to initiate a strategy centered on the mutual utility of art and business (ref. Jacobson, 1993; Behnke, 2007). Although the history of corporate collecting is relatively recent, in a short period of time a significant number of companies have acquired impressive collections (d’Arenberg, 2017). The largest corporate art collection belongs to the Deutsche Bank with 57.000 art objects (Sharf, 2012). JP Morgan today owns around

(18)

30.000 pieces of art, some are from famous artists like Andy Warhol and others are from unknown artists (Sharf, 2012). Most of the firms that started with the trend of corporate art collecting where financial institutions. However, by the mid 1990s around half of the Fortune 500 companies were collecting art (Kottasz et. al 2008). With art, it is many times the case, that it is very difficult to quantify and measure the success of the positive externalities. Therefore strategic management is required to set up a purposeful corporate art collection. A group of experts that knows the company and the art world well can help a company make strategic decisions regarding buying art works. Although the literature on corporate art collecting is very scarce, there are some researches done on CACs.

The current literature is focused on the purpose of the corporate art collection as corporate identity. Corporate Identity needs to be planned and purposeful (Alessandri, 2001), here is where the strategy of the curator comes in. Kottasz et.al (2007) identified corporate identity as the key role of corporate art collecting. In this aspect collections are used to project the core values of organizations, including facilitating internal and external marketing communications, symboling and signaling, as well as stimulating employees to think differently (Kottasz, 2008). An art collection can be a very strong way of portraying to their employees, customers and other stakeholders what the company stands for. Therefore there is a key role of the curator to choose the right art for the firm’s collection, making them accountable for their choices. This study will focus on the same artists that corporates buy from, challenging the role of corporate identity and moving towards a more conspicuous consumption of art by firms. Do firms that do not buy from the same artists represent more purely the purpose of the collection of the firm, or those that actually copy? Showing the public

(19)

a certain status of the firm, social class. More expensive artists in the collection could act as a kind of status symbol. What is usually known as conspicuous consumption (Trigg, 2001).

Different purposes of CAC that have been examined in the literature vary among firms. The Deutsche bank uses their corporate art collection to inspire their employees, and to signal corporate identity dynamics in which the firm is situated in the present (Banhke, 2007; Kottasz et.al, 2007). Their first art acquisitions were mostly on canvas or on paper, and as the bank became more digitalized, their art collection also started to become more digitalized. Trying to convey an image of innovation and dynamic adaption towards today’s digitalized world (Kottasz, 2008). The curator has an important role in potraying the firm’s standards and beliefs. This role creates a certain pressure on the curator to perform well, according to the firm’s guidelines and corporate identity. There is no literature taking into account this pressure on curators to fulfill their role correctly in the firm. This paper takes into account this pressure, or accountability, in the same way investors in the financial markets have the pressure to perform and maximize the profits of their investments (Fracassi, 2016).

Likewise, instead of corporate identity purposes, other studies emphasize corporate art collection as tools for public relations and networking (Dell’era, 2010; Behnke 2007). This purpose of corporate collecting is interesting to try and understand it because there is little to no empirical evidence of the network structures and influences in corporate art collections. This gap in the literature gives room to explore the nature of the collections, and the influence a type of networks that can trigger certain buying patterns. Several researches study the purpose of corporate art

(20)

collections but not the networks of firms that comprise CACs. Other empirical studies have studied corporate collecting from the perspective of mindfulness (Barry and Meisiek, 2010). This perspective encourages employees to have more awareness of the moment, and to notice details surrounding their environment that would otherwise be unnoticed. This in turn helps to think from a different angle and in a different way than usual. The different purposes of the CACs put different responsibilites on the strategy the curators have to take. Leading them to find ways to fulfill their task in the best way possible. Probabbly by oserving their surroundings and what other firms are doing.

H1: Curators are held accountable for the purpose of the collections of the firm.

3.2. VBCN and the Small World Network

! !

A theme in this work is that behavior, opinion, and information, broadly conceived, are more homogeneous within than between groups (Burt, 2004). The corporate art world in The Netherlands is considered to be a relatively small world compared to other open networks. The scientific origin of small-world research is now commonly associated with a research done by Pool and Kochen in 1958 (Schnettler, 2009). The phrase “small-world” which has been defined in the introduction suggests that social networks are in some sense tightly woven, full of unexpected links between individuals that seem far from one another in physical or social space (Milgram, 1969 p. 426: Milgram and Travers, 1969). The VBCN is an association of firms that collect art in The Netherlands, which already comprises a

(21)

clear boundary in this network. Allowing this study to seek reasons that the VBCN is a small-world network, which can boost the herding behavior that is being observed. Even though it is assumed that the VBCN network is a small network, in the absence of empirical data, it is difficult to know which simplifying assumptions are likely to be fruitful (Milgram, 1969). However, the assumptions that the VBCN is a small world are made because the population of this study is small, well defined and homogenous. The VBCN is defined as homogenous because all the firms that are members have something in common, which is collecting art for a firm. Additionally all firms are connected by the association, the board members.

Schnettler (2009) identifies three types of dimensions of the small-world phenomenon briefly mentioned in the introduction; the structural, process and the psychological. The structural dimension refers to the likelihood of two individuals to share common acquaintances. The psychological dimension refers to the surprise of the chance encounter. The process dimension refers to the assumption that different social processes can occur through channels of contacts. In theory this dimension encompasses studies that are focused on the knowledge about ones network and the logic behind reaching out to individuals in their social network (Schnettler, 2009). This paper will focus on the process dimension.

The main characteristics of a small world are that (1) it is cohesive, (2) clusters linked by agents who are members of multiple clusters, making large communities made out of a lot of small clusters (that have strong cliques between them) being clustered and cohesive, (3) have a short global separation and have (4) high local clustering (Uzzi and Spiro, 2005: Suvillan et.al, 2014). The VBCN (2017) fulfills these characteristics in a way because they have a short global separation; they are all in the Netherlands.

(22)

The firms belong to an association, making them clustered and cohesive. This association creates events and VIP evenings during different exhibitions and art fairs in order to gather the curators and give updates (VBCN, 2017). These situations allow common ground for the curators to obtain information from each other and communicate their new acquisitions with each other. Gallery owners can also act as nodes in the network to obtain information of curators actions. Where curators might meet with gallery owners and these transpass the information to other curators. Visible opinions are likely to be salient in cohesive groups because members are embedded in a field of interpersonal cross-pressures that encourage reciprocity and compromise (Friedkin, 1993). Hence the greater the structural cohesion of the network, the more likely they are to influence each other (Friedkin, 1993). The greater the structural cohesion the more likely they will be in frequent communication on a particular issue. Curators that talk about artists somehow could create uniformity of opinions towards the work of an artist in the network. As well as those most central curators who are admired the most, those opinions can be most prevalent in the network, influencing others choices on artworks to be bought.

There are two main views in the small world network literature. One is that a small world closes itself from new information (Friedkin, 2012; Uzzi and Spiro, 2005). Some researches explain this phenomenon as ‘The small world problem’ (Milgram, 1969; Schnettler, 2009). The other view is that small-world networks actually allow newer information inside the network (Sullivan, et al. 2014), which will be discussed shortly. This paper takes the view that a small-world network allows less information in the network. Due to the fact that at some point the small-world stagnates the amount of information coming inside the network, limiting the amount of new

(23)

suggestions by new curators about artists out there. In their research Uzzi and Spiro (2005) found that creativity was affected by the small world, and reduced input of new information at a certain point. Creativity increased to a certain point of small world network degree and then after that point it decreased. The argumentation followed in Uzzi and Spiro’s (2005) paper is that creativity is the consequence of an interactive social system of actors that, in their paper, if closed to newcomers suppress new creativity. Parallel argumentation is used in the current situation in CACs collections, where artists are recurring in different collections of the VBCN can be due to the small world in which the firms are embedded.

Somehow contrary to what Uzzi and Spiro (2005) argue that small world can lead to a smaller amount of influx of novel information absorption in a network, there is also an opposing view to this. The opposite view is taken when firms don’t have a high level of recurring artists and have a larger variety of artists in their collection. Sullivan et.al (2014) argue that then the small network scenario actually allows more access to novel information. Relative to those that are not in a small network, firms embedded in a small-world network are more likely to develop routines for searching beyond network boundaries as distant information becomes more accessible to the individuals (Sullivan et.al 2014). The current study posits that a small-world network reduces the amount of information coming inside the network, because the curators that belong to the association of the VBCN are constant at least several years in a row, reducing the rotation of new people and therefore new information in the network. The small-world problem helps to clarify and understand the purchasing behavior of curators. Those firms that have more unique artists that are not recurring in the VBCN network are assumed to follow the advantage of the small-world network. Those firms that

(24)

have a high level of recurring artists in their network are more likely to have limited influx of new information coming in the network.

There is a reduction in the entry of information about new artist in the VBCN once there is a greater ‘small-world’ network. This suggests that entry of new information is somehow restricted. The more a network exhibits characteristics of a small world, the more connected actors are to each other and connected by persons who know each other well through past collaborations or through having had past collaborations with common third parties (Uzzi and Spiro, 2005).

H2: The VBCN shows characteristics of a small-world network

3.3. Herding Behavior

!

The phrase of ‘herd behavior’ was first introduced by Trotter in 1908 (Raafat et.al 2009). A theory called the ‘herd instinct’, an innate psychological dependency of humans on their social group (Raafat et.al 2009). An individual, in the financial sector, is said to herd if he or she decides to invest when he or she finds that others have decided to do so (Bikhchandani and Sharma, 2000). According to this definition herding in the context of corporate art collections can result from an obvious intent of curators to copy the behavior of other curators, regarding the choice to invest in an artist artwork for their collection. This paper uses rational herding behavior as an explanation for the observed herding behavior. Empirically, “rational” herding occurs when decisions makers suppress their private information because making a bad decision is less costly when others make the same decision (Scharfstein and Stein, 1990: Cont, and Bouchaud, 2000). As curators may feel accountable for their role in a

(25)

firm as curators, and prefferably act in a risk-averse manner they may feel safer when they mimic others. This paper assumes that curators profits depend on their performance in relation to some reference benchmark of quality and may mimic this benchmark to reduce the risk of losing part of their income or their job (Maug and Naik, 1995). The problem is, that in the primary art market, there is no clear benchmark of quality.

The study of herding has an extensive history, not only in financial economics. This theory has been studied in two perspectives of analysis; the pattern-based and transmission-based views. The transmission perspective seeks to unify and identify mechanisms of transfer of information in herding by concentrating on the role of cognitive and affective components, particularly, the effortless human capacity to mimic the behavior of others (Raafat et. al, 2009). The pattern-based view of herding is in terms of patterns of interaction among the agents, which are modeled as simple imitators, focused on patterns and not on people (Raafat et.al, 2009). The most intuitively understandable pattern-based perspective of herding is found in social network analysis (Raafat et.al, 2009). This paper will focus on the pattern-based approach to studying herding behavior and will build on this theory. This view sees agents as simple, well-defined properties and modes of interactions that yield herding. Langley, et.al (2014) examine the patterns of herding behavior in an online setting. They explore the patterns of mass consumer behavior that leads to herding. The patterns they found are regarding the speed of contagion and the number on individuals that express this behavior. Similarly the current research on herding in CACs will look at the percentage of the sample that expresses the same buying behavior.

(26)

Recent work has found that consumers’ decisions about cultural products are influenced by popularity of the products from a variety of sources; rankings, awards, recommendations, reviews, media attention, etc. Curators in our study are the consumers of art in the primary art market; they are regarded in this study as expert buyers. Some may be more experienced than others. Curators may engage in this behavior because these expert buyers may benefit from coordinating their choices of buying an artwork, in the immensity of choices, by buying the same things as others (Adler, 1985). Observational learning is another term used for herding behavior. The theory of observational learning has much to offer to economics and business strategy because it helps explain some otherwise puzzling phenomena about human behavior (Bikhchandani, et.al, 1998). In this paper, the recurring artworks of artists in several different firms collections will be considered to be a rational form of herding behavior that can be explained by certain network structures and market conditions, regarding perceived signals of quality.

Another way to explain the herding phenomenon could be by understanding the information cascades that illustrate the rational approach that economists have applied to herding, in contrast to that of social psychologists, which are focused more on contagion and epidemics (Raafat et.al, 2009). Information Cascades occur when a person observes the actions of others and then, despite possible contradictions in their own private information signals, engages in the same acts as the others (Raafat et.al, 2009; Bikhchandani, Hirshleifer and Welch 1998: Welch, 1996). The underlying assumptions of information cascades are that agents choose in sequence and are influenced by rational consideration of the observed choices of earlier agents (Raafat et.al, 2009; Bikhchandani, Hirshleifer and Welch 1998). This sequence of information

(27)

is hard to track down, therefore the pattern-based approach to explain herding is more appropriate for this study. So contrary to the rational herding this paper has been arguing up until now, there is also a possibility that curators make decisions more unconsciously, it can be the case that a curator buys the same art as their competitor, regarding industry, however withtout actually knowing that their competitor also posses it. There are researchers like Prechter and Parker (2007) that argue that in finance, where there is uncertainty about valuations induces an un-councious, non-rational herding. Which in turn lead to financial fluctuations. So despite literature believing that are agents that are involved in herding act rationally there are also counter arguments that believe that this is not true. Their research argues completely the contrary, that in conditions of certinaty, people tend to herd consciously, and under conditions of uncertainty people tend to herd unconsciously (Prechter and Parker, 2007). Most schools of thought have emphasized the rationality of man regarded imitation as a purposeful (conscious, rational) learning strategy, and ignored the instinct theories (Prechter and Parker, 2007; ref. Neumann, 1993 p.116). Even though neither reason nor instincts offer a full explanation for curator’s motivation to buy artworks, this study is a step forward to understanding this human social behavior in economics. Especially in a market like the primary art market that isn’t considered the usual utilitarian goods market (Velthuis, 2003; Caves, 2000) which remains underexplored, or the financial market where there is still room to research how to predict certain aspects of social behavior (Prechter and Parker, 2007).

The key issue that this paper wants to explain in the theoretical framework how individuals, the curators, determine which alternative is better (Bikhchandani, et.al 1998). Each individual can decide by direct analysis of the alternatives. This can be

(28)

costly and time consuming (Bikhchandani, et.al 1998). Normally economic theories emerge from the neoclassical perspective of perfect. However in the field of information economics it has been widely recognized that information is imperfect, obtaining information is costly, and that there are important asymmetries of information, and the extent of information asymmetries is affected by actions of firms and individuals (Stiglitz, 2000; 2002). It is assumed in this paper that curators are not able to seek all the information about all the artists in the entire world and make the best decisions regarding all options. They are limited to the information they have and the information that reaches them or they seek by asking others. A good alternative is to rely on the information, signals and actions of others. This influence can be exerted in the form of direct communication and discussion with, or observing others (Bikhchandani, et.al 1998). Through intermediaries and links in the network of the VBCN this is possible. Paradoxically, the ability to learn by observing others actions can make the decisions of followers noisier by reducing their incentives to search for information themselves. This noise means that there is a possibility that a self-fulfilling prophecy occurs. Which would mean that a false definition of the situation evokes a new behavior, which makes the original false belief come true (Salganik and Watts, 2008). Small worlds bridge ties between otherwise distant neighborhoods that have been shown to promote the diffusion of information, innovations, and rumours are shown to inhibit cascades of false enforcement (Centola and Macy, 2007; ref. Granovetter 1973). The reason behind this is because by not searching for information themselves they rely on what others regard as high quality.

H3: Curators of corporate art collections tend to acquire artworks from

(29)

network of the VBCN

!

3.4. CACs and the Primary Art Market

The previous section has highlighted the CACs and and network setting that can be a possible explanation to comprehend the herding behavior of curators in the VBCN. There are several reasons why organizations would buy the same artists and engage in the so called herd behavior, however there are two salient reasons that can lead curators to herd. These are (1) accountability issues, which have been discussed, in the previous chapter and (2) to overcome uncertainty about quality. The first reason, to recap, regards curators as experts in buying art for their organizations and they may want to diminish the perceived risk (Simon, 1972). These can be regarded as accountability issues. These accountability issues stem from a curator not understanding the meaning of quality regarding art. In rational herding models agents consciously and deliberately consider the information in the signals of others to make decisions (Raafat et.al, 2009). The second reason focuses on that the primary art market is very broad, and is characterized by having an over supply of artists, which leads to an uncertainty in demand (Menger, 1999) and a market where there are no objective measures for quality evaluations (Menger, 1999; Caves, 2000).

The art market, which is the context of this study, focuses on two main players to explain the unequal diffusion of artists. On the one hand you have the artists, which are the producers of artworks bought by the organizations that collect art. On the other hand you have expert buyers, in this case the curators that select the artists artworks they want in their collections. The first section of this chapter will discuss in detail the characteristics of the art market. Then two possible suggestions to overcome

(30)

uncertainty in this market and act risk-averse will be discussed. The first suggestion is by looking at the most experienced curators; those curators that are less experienced are predicted to herd more than the most experienced ones. This research expects a negative relationship of experience on herding. The second suggestion to overcome uncertainty is by looking at the closest competitors; those in the same industry are expected to have a positive relationship on herding.

3.4. 1 Current state of the primary art market !

There are certain characteristics that dominate the creative industries market. These have been pointed out by various researches. The first characteristic of the creative industries is that there is an oversupply of creative producers (Menger, 1999). Several reasons account for this oversupply, one reason includes the increased support for cultural industries by governments, which increased employment opportunities in this sector, creating more need and demand for these producers (Menger, 1999). Another reason is the overall increase in income per capita which allows to demand more cultural products (Ghoshal and Moran, 1996), and in turn increases the demand for creative producers in the market, and to compensate this increase demand, the supply or creative individuals also increases. The third reason is the low barriers to entry (Menger, 1999). Lastly there is an oversupply of artists when the market became a free art market with a more flexible and riskier basis for open competition, involving more parties in the process such as dealers, critics, buyers and gatekeepers (Menger, 1999; Wijnberg, 1997). This oversupply is important in understanding the herd behavior because there is a vast amount of choice curators face. There is so much information, that some quality signals can help to sort this information out.

(31)

Those artists that have more impact in between the entire mass of artists are those that have more persuasive attraction to be bought, because they have more attention. The impact of an artist can be the reason why it stands out from the entire supply of artists, and influence a curator to buy their work.

Next to the oversupply of artists there is another characteristic that dominates the art market. Creative industries are also characterized by uncertain demand. Which Caves (2000) refers to the ‘Nobody Knows’ fundamental characteristic of these industries. The ‘nobody knows’ aspect relates to the lack of quality signals about all the creative suppliers. Artists in this case, which are our creative suppliers or producers, make decisions regarding their products quality, however they are not able to predict the audience perception of quality (Caves, 2000). Hirsch (1972) points out that first of all there is no precise formula or ingredients to best sell the creative product. Hirsch (1972) describes cultural products as non-utilitarian, and serve more as an expressive and aesthetic products rather than useful. This makes it difficult to have a quality definition of the artist’s products. In turn this makes it difficult for curators to choose one artist over another for their collections. This aspect of the primary art market also creates more attention for those artists that are more popular and understood. Again incentivizing curators to copy others behaviours.

3.4.2. Quality signals of artists and third parties !

Stiglitz (2002) views markets from another perspective, in the light of quality signals to consumers through producer’s actions. There are two types of market imperfections, competition and information. One of the main assumptions of Stiligtz (2002) research is that there is asymmetry in the information in the markets. Any buyer or consumer has different amounts of information about quality of products

(32)

(Stiligtz, 2002), this is what creates the assymetry. This can depend on the level of expertise and specialized knowledge each individual has. There are some arguments that allow overcoming the asymmetric information; such as conveying information through actions (Stanglitz, 2002). The fact that these actions may convey information affects behavior (Stanglitz, 2002). In his paper he argues that the fact that actions convey information leads people to alter their behavior. Imperfect information and information asymmetry both combine two important effects on the recurring artworks found in collections. Imperfect information allows some curators to observe the actions of others, and information asymmetry means that some curators seem to have more information about artist’s careers than other curators. This is an incentive for those experts to be followed by the less expert curators.

This uncertainty is caused by different consumer taste preferences, but mostly by the different criteria gatekeepers use in selecting cultural items to be awarded (Hirsch, 1972). There is no precise definition of what is good quality and what is bad quality in the contemporary art market. However there are certain processes in these creative industries and signals that can smoothen these information uncertainties in the market. The characteristics of the market mentioned above hinder the access to information about quality of products in the market. Therefore curators are assumed to have difficulty choosing artists artworks. Certain quality signals can act as strong incentive to buy certain artists over others. These quality signals can come from different sources. One example of smoothing information in the creative industries are the gatekeepers (Wijnberg, 1997). Gatekeepers grind information about the abundance products information about creative products through awards (Dempster, 2006; Ginsburgh, 2003), dealers (Velthuis, 2003) reviews, critics (Schrum, 1991) and

(33)

evaluations about products and producers of these creative products. These gatekeepers are very important in the reception of quality of cultural products (Hirsch, 1972).

There are other aspects that can lower the uncertain characteristics of the arts market and can signal quality to curators. There are two main sources of quality, those of the artists and those of third party sources (Kackovic, 2016). Signals of artist’s producers provide quality information to potential buyers and decisions makers, the curators (Kackovic, 2016). However the signals from the artists, the producers of artworks, are considered as subjective information, because this quality information comes from the artists themselves. It has also been proven that artists are not the best selectors when it comes to evaluating their own work (Berg, 2016). Third party signals are more believable than the producer’s subjective opinion about their work. Third party sources can range from educational background, reputation awards to prestigious galleries that sell these artist artworks. Another quality signal can even be another firm in the same industry buying from the same artist. First of all, the chance that information arrives to you as a firm in that industry is much higher than if you are in a completely different industry. Secondly if they are doing it, they might think “we must be doing something wrong”. Notwithstanding these sources also bestow different levels of credibility. As Kackovic (2016) puts it, the strength of the signal depends on the credibility of the third party source. Artists which are on the international art fairs galleries on display in Art Basel or Tefaf for example, have more chances of being more popular among corporate art collections than those that are in local exhibitions of a small town in the Netherlands.

(34)

3.4.3. Overcoming the uncertainty of the primary art market by looking at firms in the same industry and expertise of the curators.

Curators and firm industry

!

Like any other market, in the contemporary art market, there exists an overload of information creating uncertainty about quality of a product. Nelson (1970) argues that there are limitations to consumer information about quality of products in any market. Information about quality of products is more difficult to obtain than prices for example (Nelson, 1970). Which makes the buyer more reliant on observing other behaviors and use these as simple quality signals (Stiglitz, 2002). As mentioned before, there is no clear benchmark of quality. One way to compare your collections quality with anothers firm quality is by observing what the other firms in the same industry are doing.

Inter-organizational imitation is a useful mechanism for understanding curator’s actions that result in the rise of a dominant strategy. The strategy of buying an artwork from the same artist as the rest of the curators in the same industry. Since it is the safest decision. Firms that face uncertainty over strategic alternatives of buying artworks rely then on industry comparisons to develop their strategy (Low and Abrahamson, 1997; Abrahamson, 1991; Xia, Tan and Tan, 2008). On a firms level then firms that are in the same industry are expected to herd more with each other than those that are in distant type of industries. They are more likely to observe the behavior of close competitors than firms in other industries. In the same way that Fracassi (2016) states that firms that are in the same industry are more likely to invest similar amounts than firms in a different industry, this paper argues that firms that are

(35)

in the same industry are more likely to invest in the same artist than those firms that are more distant to their own characteristics and market they are not involved in.

In a way an artist need to give out signals to surpass a legitimacy threshold in order to even be considered by curators. So in order to limit the quality uncertainty of the primary art market it would seem like a good idea to observe what the neighbor is doing. In the same way Low and Abrahamson (1997) state entrepreneurs, or newbees, in a sense need to be understood by the market before consumers consider them. The way in which new artists can give out signals of quality, can also be explained by

movements and bandwagons. A movement is the organizational process by which

new organizational forms are created; bandwagons are organizational processes that seek to exploit the potential of a new legitimized form (Low and Abrahamson, 1997). In the context of artists, curators, and corporate art collection the parallel argument that can support this thought can easily be explained. In the same way a movement is the process at which an organization form is created by allowing consumers understand a new form of organization, artists bought repeatedly have shown that consumers have understood their work. Buying from an artist that was unknown, which later on has been bought by more firms, allows this artist to constitute a movement. On a deeper level, an artist that has been bought by more firms, and those firms being in the same industry can create an even closer association of quality. Since it is embedded in a circle of firms it is familiar with. This artist is increasingly showing that it is legitimate. Others have understood his or her work, it has given a signal of quality, and therefore it is safe to buy from this artist for a corporate collection.

(36)

Bandwagon effects can also be an argument for the logic that firms are engaging in the same buying behavior because they are in the same industry. The bandwagon would be able to explain why some artists are being bought more than others. There are some firms that want to exploit the benefits of acquiring an artist that has been legitimized. This sign lowers the amount of information that needs to be sought in the primary art market. If a firm exploits this new legitimized artist, it means that it has obtained signals that it is popular and successful among other firms that are in the same industry. Which in turn can cause firms to be scared to miss out on something that is regarded as qualitative. In order not to miss the bandwagon they engage in the same buying pattern. Buying that artist that has been bought by the neighboring firm. If, for instance, a firm in an industry buys an artwork for their collection from an artist, and this demand drives up the qualitative value of their collection, then other collections seeking high quality in their collection will follow these firms in the same industry.

These two arguments can be understood as driving forces for purchasing the same artist to overcome the excess of information and quality uncertainty in the primary art market. In order to lower this uncertainty firms therefore herd towards the new ‘fad’ of the industry they are embedded in. Therefore this paper argues that firms that are in the same industry have more chances of herding with each other than in between firms. Choi and Sias (2009) examine whether investors follow each other into and out the same industries. The fads argument proposes that investors may herd to industries simply because those firms in the industry have become more popular. Following the argument of boundedly rational behavior, a firm in the same industry that has bought an artists artwork, is a driver for another firm that is collecting in the same industry, to

(37)

purchase artwork from the same artist. By testing this hypothesis we investigate the diversification of artists artworks while taking into account a firms industry specific component. This paper expects a positive influence of industry type on herding. Firms within the same industry have a positive influence on the observed herding behavior. Those that have a negative beta coefficient will seem to be firms that don’t herd within the same industry.

H4: Curators of corporate art collections tend to acquire artworks from other

artists that have previously been acquired by curators that are in the same

industry

Curator’s expertise and bounded rationality

!

Up until the 1990s top managers almost always administered their own corporate art collections (Barry and Meisiek, 2010). Since then, most collections have been handed over to professional corporate curators with legitimacy and networks in the art world (Barry and Meisiek, 2010). These curators try to collect art that meet organizational needs and managerial expectations that match the work processes of the organization (Barry and Meisiek, 2010; Behnke 2007). Those curators with larger knowledge and a larger network might influence those junior professional curators with less expertise in the field that just entered CAC world. Leading those with less expertise and specialized knowledge to buy from the same artists the more expert curators buy from. There are two types of curators. Ones that are experts and other that are characterized as non-experts. The effects of the signals differentiate between what type of curator is faced with these signals (Kackovic, 2016). Expert buyers are those that are highly skilled and have specialized knowledge about the art market. These have a professional experience, for example full time curators and

(38)

conservators. Compared to those curators that work part time in a non-profit organization and do it as a hobby. These curators are less informed buyers and lack the familiarity and expert knowledge about the products and or the producers they are considering to purchase (Kackovic, 2016). The observed actions of the expert curators can signal a quality to the non-experts and therefore reduce the risk of choosing between infinite alternatives.

As mentioned earlier in this paper, obtaining information is time consuming. Due to the imperfect, incomplete, and the overload of information it can be argued that the effect of these signals actually result in bounded rationality decisions (Simon, 1972, 1991). Rationality means that a certain type of behavior is appropriate to achieve a certain goal, within the limits imposed and constraints (Simon, 1972). However, there are some limits to rational behavior that is why it is bounded. Bounded rationality refers to the limited amount of time and money a person has to consider all the available options, therefore it makes a decision with the information that it is constrained with. Curators don’t have enough information about the entire art market because strategists assume that firms are bloodedly rational (Phelan and Lewis, 2000). Therefore this paper argues along the same lines as Kackovic (2016) and assumes that one way curators overcome these market characteristics of quality uncertainty of the contemporary art market is by using this signals to diminish the overload of information out there. These signals are those of publicly observable events or actions that convey information about artist’s unobservable quality (ref. Spence 2002, 1974, 1973; Kackovic, 2016).

(39)

As discussed in this chapter in the previous paragraphs there is some room to believe that certain quality signals allow firms to be more inclined towards buying an artist’s artwork over another one (Spence, 1974). This paper uses the argument and assumes that artists that have some kind of title or performance-based recognition, like an award, or are more demanded by firms, so called publicly observable signals, create self-reinforcing loops of demand. This explains the phenomenon of unequal diffusion of artists artworks and more chances of being selected by firms because it assures some degree of quality to the firm. The assumption this paper supports is that different quality signals from the artists as well as well as the level of expertise from the buyers can explain the herding behavior of curators. From the artists the legitimacy that can be driven by different sources such as press, media, being owned by galleries, etc. From the expert buyers point of view, those that have more expertise, are more prone to be followed by other curators that are less expert buyers. In this way the observable signals portrayed in the action of expert curators can lead to other non-expert curators to follow the same buying decision. All these signals can project the popularity of the artist and create more demand for an artist’s artwork in the collection. Due to bounded rationality curators rely on quality signals from gatekeepers and other third party sources when making buying decisions. These arguments provide and explanation and some insights into understanding the incentives curators face to herd.

Theoretical literature suggests other reasons to herd, including reputational herding (Villatoro, 2009). Chevalier and Ellison (1999) indicate that there has been little empirical work documenting how career concerns may affect managerial behavior in any industry. The career concern curators have is the accountability they have towards