The Price of Purity
Brokerage as Consecration
in the Market for Modern Art
Fabien Accominotti∗
Columbia University
This article proposes a structural approach to consecration, and uses
this approach to solve an empirical puzzle in the sociology of markets. Markets for novel and unique goods are often seen as privileged
settings for the powerful influence of market intermediaries. Yet economic sociologists have repeatedly failed to observe any impact of art
market brokers on the value of the artists they distribute. This puzzling finding, I argue, arises from a misconception of what brokerage
does to economic products. While market intermediation is usually
thought to act through two social processes of valuation – certification, or the signaling of underlying quality, and qualification, or the
establishment of specific quality criteria – I suggest that it also influences value through consecration, or the conferral of relational purity.
This approach fills a gap in economic sociology, which lacks a distinct
definition of consecration. It also suggests a network-based strategy
for capturing consecration empirically. I finally show that brokerage
as consecration, not certification or qualification, is indeed how art
market intermediaries shape the value of their artists. The empirical
focus is on the market for modern art in early twentieth-century Paris
– a setting chosen both for its historical significance and its ability to
magnify how markets, as social environments, bear on the economic
worth of things.
∗
I am particularly grateful to Peter Bearman and Pierre-Michel Menger for their support, and to Jens Beckert, Diana Crane, Thierry Dufrêne, Victor Ginsburgh, Olav Velthuis
and Harrison White for their insights and suggestions. For helpful comments on an earlier
draft I also thank Paul DiMaggio, Shamus Khan, Uri Shwed and David Stark, as well
as participants to the 105th Annual Meeting of the American Sociological Association in
Atlanta, to the 22nd Conference of the Society for the Advancement of Socio-Economics
in Philadelphia, to the 2012 Max Planck Summer Conference at Northwestern University,
and to seminars at Columbia University and CNRS. Mailing address: Fabien Accominotti,
Department of Sociology, Columbia University, 501 Knox Hall, 606 West 122nd Street, New
York, NY 10027. Email: fa2224@columbia.edu.
1
“Yet what happens here has its counterpart in our own society. A painting by a famous
artist sells for many thousands of dollars at a certain time because we have agreed in
effect to reduce to the common denominator of economics all products and all goods,
even those products which nowadays are increasingly acquiring religious value.”
Louis Dumont, Homo Hierarchicus: An Essay on the Caste System
“Nothing but their heterogeneity is left to define
the relation between the sacred and the profane.”
Emile Durkheim, The Elementary Forms of Religious Life
Introduction
A paradox awaits the sociologist venturing into the study of art markets.
Markets for unique goods are usually seen as privileged settings for the powerful influence of market intermediaries: when quality is uncertain, or when
it lacks definition altogether, market brokers can play a crucial role in signaling or specifying it, thereby ultimately shaping the prices consumers are
willing to pay for products. Products, meanwhile, do not get much more
unique than in the market for contemporary art. Yet oddly enough economic
sociologists have repeatedly failed to identify any value shaping mechanism
involving art market intermediaries. The reputation of art dealers, for example, does not seem to bear any noticeable impact on the prices of their
artists. Nor do their promotional campaigns, their experience in business, or
their commitment to more highbrow forms of artistic production. And this
gets even more perplexing, as dealers, artists, and sociologists themselves
nonetheless continue to regard intermediaries as the makers or breakers of
artistic careers.
The present article offers a solution to this puzzle. To this end, it advances a sociology of economic valuation. Specifically, it draws an analytical
distinction between three types of processes – certification, qualification, and
consecration – whereby markets, as social environments, can affect the value
of marketed things. It also suggests that while both certification and consecration have received considerable attention in the literature on markets,
economic sociology lacks a distinct approach to consecration. This, I argue,
is the reason behind the paradox outlined above. At its core, this article therefore offers a theory of consecration, and a strategy for capturing
2
it empirically. To anticipate, it approaches consecration as the operation
whereby intermediaries confer relational purity to products in a web of affiliations (Breiger, 1974). This relational form of purity departs from the
ritual purity one can acquire by complying with culturally scripted conducts
or established categories (Dumont, 1970; Douglas, 1966; Abbott, 1981; Zuckerman, 1999, 2004).
Though the idea of consecration obviously has religious undertones, the
approach I propose insists on consecration’s structural features, and makes
it identifiable regardless of the mundanity of the setting. As an illustration,
the article eventually goes back to the initial problem to show that brokerage
as consecration, not certification or qualification, is how art market intermediaries shape the value of their artists. The empirical focus is on commercial
galleries distributing modern art in early twentieth-century Paris. Yet the
article helps reflect more broadly on how the social wiring of markets bears
on the economic worth of things.
Social processes of valuation in markets
Identifying the social underpinnings of market valuation has long been a key
enterprise of economic sociology – and understandably so. The effort, after
all, addresses head-on the radiant core of economics, promising to show just
how ingrained market value is with social matters. It also involves reflecting
on the non-intrinsic, and therefore perhaps questionable, origins of value
inequalities across economic entities: there is indeed something troubling to
the fact that individuals, firms, or goods of comparable ability or quality
may, because of differing social backgrounds or embeddedness, be valued
differently.
Speaking to these various concerns, social science essentially has two
frameworks within which to approach the social drivers of market value.
The first, which one can term the quality certification or revelation framework, builds on the premise that the quality of products up for exchange
in economic life is often uncertain. When this is the case, other product
characteristics may serve as informational cues signaling their underlying
quality. Such characteristics can be an individual’s education, the past performance of a firm, or the praise a product receives from experts. They
need not be social in a very strong sense – and actually early proponents of
this framework can be found within economics itself (e.g. Spence, 1974). In
more recent years, however, the idea that the relational and cultural wiring
of markets can also work to alleviate quality uncertainty has enriched this
3
approach with a deeper sociological twist. In blurry economic settings, it
has thus been found, the popularity of firms with business partners, or the
status they acquire from displays of deference by competitors, can help thirdparty customers assess the quality of their products (Podolny, 1993, 2005;
Benjamin and Podolny, 1999). Likewise, personal ties underlying market
transactions can entail an element of trust, or the disclosure of private information regarding product quality. Customers as a result may prefer to
pay a premium for products acquired through socially embedded exchange
– so long as this premium does not exceed the costs they would incur to
collect equivalent information otherwise (e.g. Geertz, 1978; Uzzi, 1999; Uzzi
and Lancaster, 2004; Velthuis, 2005; see also DiMaggio and Louch, 1998).
The kind of informational cues certification processes breed on may also be
more cultural in nature. Higher prices, for example, are sometimes interpreted by customers as signals of greater quality – this is notably the case
with contemporary art (Moulin, 1987; Velthuis, 2005). In a more subtle way,
culturally scripted pricing rules can also function to alleviate the sense of
uncertainty associated with certain products (Velthuis, 2003). And cultural
categories can interact with the certifying work of experts to influence market value, as when entrenched classification systems divert the attention of
market analysts from certain firms in an industry (e.g. Zuckerman, 1999,
2004).
The certification framework often assumes the correspondence between
quality and quality signals to be loose. Informational cues, that is, do not
necessarily reflect or reveal quality with great accuracy. This has an important implication, as it helps explain why discrepancies in market value
sometimes depart from actual quality differences between economic entities.
Yet as the mere possibility of this implication makes clear, a basic tenet of
the certification framework is that there exists relative agreement as to what
constitutes underlying quality.
This is not the case in the second framework, which one may refer to
as the quality definition or qualification framework. Most central to this
other approach is the notion that the nature of quality can itself be elusive,
and should not be taken for granted in the first place. The adoption of
certain quality criteria, as a consequence, can have a profound impact on
the ultimate value of things. This is perhaps no more obvious than when
markets emerge for products which once fell outside the scope of commercial
exchange. When it comes to giving monetary value to nature, life, or cadavers for example, the issue is not only to overcome the moral preventions that
previously kept them out of the market sphere. Also at stake is the question of what, in these things, will be deemed valuable, or to put it starkly,
4
of what will count (Zelizer, 1979, 1981; Fourcade, 2011; Anteby, 2010). In
such markets and others, qualification processes bring answers to precisely
that question (e.g. Espeland and Sauder, 2007; Stark, 2009; Beckert and
Aspers, 2011). By collectively defining the valuable, they shape economic
preferences and quality beliefs, thereby eventually influencing the value of
products.
Shared understandings of what constitutes quality can thus emerge from
interactions between market participants. A startling example is Moshe
Adler’s account of superstardom in cultural markets (Adler, 1985). The
utility consumers derive from a cultural product, Adler argues, increases
when they can discuss it with one another and thus acquire interpretive
frames that enhance their appreciation of the product at stake. In Adler’s
view, this explains why the demand for performers with already large audiences is more likely to grow even higher over time. In a similar vein, the
quality of contemporary art is often constructed through efforts by various
market intermediaries to elucidate it to potential collectors. But the social
definition of quality can also be more contentious, when fierce struggles oppose various quality principles and resolve in the adoption of the one with
the most powerful proponents. Pierre Bourdieu, in his classic account of the
market for symbolic goods, offers a distinct elaboration of the latter scenario
(Bourdieu, 1993).
As a quick overview of the examples above suggests, markets for cultural
products have often been used as strategic research sites for identifying social valuation processes. Whether the market for wine, art or music, they
indeed typically involve unique or novel goods, the quality of which is often
uncertain, or needs to be collectively constructed. It is therefore intriguing
that several efforts to observe the impact of such processes on value in what
is perhaps the most emblematic of these markets, namely the market for
contemporary art, have recently failed.
Art market brokers and economic value: previous
research and ongoing puzzles
The art market often has observers puzzled over the prices works of art sell
for. This is certainly the case when auction sales occasionally beat record
prices, grabbing headlines. Yet it is also true, in a more subtle though
perhaps more revealing way, when a piece that was virtually worthless a
couple of years earlier later goes for a noticeable amount.
That previously worthless art later sells for high prices is often accounted
5
for by the support it receives from various market intermediaries (e.g. Becker,
1982; Mulkay and Chaplin, 1982; Moulin, 1987; Velthuis, 2005). And what
intermediaries do to bolster the value of their products has precisely been
analyzed both in terms of certification and qualification.
The mere promotion of their artists by intermediaries, typically acting
through qualification – or the construction of collector’s preferences – is thus
an obvious candidate to explain their impact on prices. In addition, because
the quality of contemporary art is often uncertain, intermediaries can provide customers with the kind of quality signals needed to make consumption
decisions. These signals may of course come from expert prescribers – such
as art critics – or various judgment devices – such as guides or rankings
(e.g. Hatchuel, 1995; Bonus and Ronte, 1997; Karpik, 2010). Yet market
intermediaries themselves can also act to certify the quality of the products
they distribute. Studying the role of talent agencies in the Hollywood film
industry, Bielby and Bielby (1999) thus found that representation by an
elite agency helps authenticate the quality of screenwriters. Writers with
elite representation are therefore more likely to find employment, and earn
considerably more on average than equally accomplished writers with nonelite representation. In similar fashion, one may anticipate the reputation
of a gallery to convey positive signals regarding the quality of its artists,
thereby enhancing their market value over and above other indicators of
artistic achievement (Beckert and Rössel, 2004).1
The problem
Interestingly, the role of intermediaries in shaping the value of contemporary
art has been explored in two recent studies in economic sociology (Rengers
and Velthuis, 2002; Beckert and Rössel, 2004; see also Velthuis, 2005). And
these studies have come to fairly surprising conclusions, as neither of them
could identify any influence of intermediaries on the prices of their artists.
Although their empirical settings varied significantly, both articles sought
to measure the impact of the certification and qualification work of galleries
on the prices of their artists. Beckert and Rössel focused on the certification aspect of the problem, and operationalized the certification power of a
gallery through its reputation, as assessed by a panel of experts. Rengers
1
One third mechanism may actually account for the impact of intermediaries on the
prices of their artists: they could simply provide artists with better work conditions and
thus unleash their potential to produce more valuable art. While this hypothesis is not
directly addressed in this article, it is also at the core of Bielby and Bielby’s findings on
talent agencies.
6
and Velthuis in contrast measured reputation as the ability of a gallery to
have its artists represented by contemporary art museums. In both cases,
however, reputation was meant to capture the perceived ability of a gallery
to discern artistic quality. Rengers and Velthuis in addition also looked at
the qualification work of intermediaries, or in other words at their ability to
shape customers’ preferences. They did so through a measure of the number
of artists a gallery represents. The lower that number indeed, the more likely
each artist is to benefit from the promotional efforts of her gallery. The two
studies finally measured prices differently: as selling prices in galleries for
Rengers and Velthuis; both in galleries and at auctions for Beckert and Rössel. In the latter case, and as far as the effect of gallery reputation on prices
was concerned, the findings were consistent across measures of prices.2
As a matter of fact, neither article could isolate a specific impact of
galleries on the prices of their artists. Once various characteristics of artists
were adjusted for (in particular their career length and assessed quality,
or expert recognition, as measured through the consensus of art critics)
the mere reputation of their market intermediaries brought no significant
premium in their prices. More precisely, Beckert and Rössel observed that,
if anything, gallery reputation had a negative impact on prices, whether
measured in galleries or at auctions (Beckert and Rössel, 2004, 44-45).3
Rengers and Velthuis did not find a significant effect of gallery reputation
either. Actually, among all characteristics pertaining to galleries, only age
– or experience in business – had a small, positive influence on the prices of
artists. Rengers and Velthuis thus concluded that “apparently, galleries are
not able to add economic value themselves”. In other words, “the fact that
[certain] galleries sell [more] expensive works has more to do with the artists
they represent than with their own characteristics” (Rengers and Velthuis,
2002, 18-24).
These findings are doubly puzzling. On the one hand they indeed challenge our sociological understanding of the role intermediaries play in uncertain markets. More importantly, however, they also contradict the vision
that gallery-owners and artists themselves have of the operation of the art
market. In the remainder of his account of the contemporary art world,
2
Beckert and Rössel’s gallery prices were collected in a number of Berlin and Leipzig
galleries in the early 2000s. Rengers and Velthuis’ prices instead came from a much more
comprehensive database, reporting the actual selling price of over 16,000 works of art by
2,400 artists in 230 Dutch galleries between 1992 and 1998. As far as gallery prices are
concerned, then, the findings by Rengers and Velthuis are certainly more trustworthy.
3
This impact was not statistically significant, however, and the authors do not further
comment on it.
7
Velthuis (2005) thus shows that the ability to enhance artists’ prices is actually key to how dealers judge their performance in that market. Yet as
evidenced by the results above, he does not clearly identify where that ability
comes from.
Can we reconcile the observed lack of influence of galleries on prices with
the notion that market intermediaries act as certifiers and qualifiers of the
products they distribute? Several explanations in cultural economics have
sought to go down this path. They are surveyed critically below. As they are
found wanting, the next section introduces a different mechanism whereby
certain galleries do influence the value of artists – in line with the vision of
art market actors.
Candidate explanations
In a highly stylized, game-theoretical model of the primary art market,
Schönfeld and Reinstaller (2007) thus suggest that a gallery’s higher reputation should have a negative impact on the prices of its artists. In order
to increase their market share and profit, they argue, galleries could indeed
seek to attract customers by cutting prices. Rival galleries however would
then likely react by lowering their own prices in turn, and ultimately this
undercutting spiral could generate a “signal of instability increas[ing] the
uncertainty in the market and undermin[ing] its proper functioning. Therefore, according to Schönfeld and Reinstaller, it is a good strategy for galleries
to set prices [from the outset] in a way that does not allow their rivals to
profitably undercut them” (Schönfeld and Reinstaller, 2007, 146). At this
“undercut proof equilibrium”, the authors eventually show, the prices of
high reputation dealers should be lower than those of their low reputation
counterparts.4
In a slightly different vein, one could explain the puzzling relationship between prices and gallery reputation by building on Podolny (1993)’s account
of competition in the investment-banking market. Similar to high-status
banks, who can obtain their inputs at lower costs, reputed intermediaries
may also be able to put less money on the table to secure access to the work
of a given artist. They could then pass on these lower costs to the prices
they charge to customers – thereby profitably enhancing their market share.
One would thus predict a negative impact of the reputation of galleries on
4
That price variations can hurt the functioning of the market by increasing consumer
uncertainty is also central to Velthuis’s account of dealers’ pricing scripts (Velthuis, 2003).
Actually Schönfeld and Reinstaller present their model as an economically grounded explanation for such scripts.
8
the prices of their artists.
These explanations nonetheless have two major weaknesses. They first
suggest that the prices of products should negatively correlate with the reputation of market intermediaries. They are therefore relatively ill-suited to
the findings by Rengers and Velthuis or Beckert and Rössel that the reputation of galleries has no significant impact on the prices of their artists. In
addition, they ultimately rest on the reintroduction of price competition in
the primary art market. This puts them at odds with both qualitative and
quantitative accounts suggesting the existence of a downward price rigidity
in that market – a reluctance of dealers, in other words, to decrease prices –
and portraying dealers as price, rather than profit, maximizers (Velthuis,
2005; Hutter et al., 2007).
Moving forward
Contrary to these various attempts, this article does not try to explain the
lack of influence of galleries on prices. Instead, it further specifies the role
they play, arguing that earlier studies did not fully capture what dealers do
for their artists. More precisely, I suggest that brokerage can enhance the
economic worth of products by positioning them relative to one another in a
web of status-conferring affiliations. The process I propose thus goes further
than the previous two in disconnecting the social aspect of value from the
intrinsic properties of economic entities. In contrast to both certification
and qualification, where value emerges from the interplay of social factors
with product features, here the mere position of products in a web of affiliations directly translates into economic value. To put it in Dewey’s terms,
social embeddedness is no longer a means for the evaluation or appraisal of
economic entities. Rather, it becomes an object of prizing (Dewey, 1939).
That the relational status of economic entities can bear on their economic worth is hardly a new idea. In fact, classic work in the sociology
of markets has shown how the status firms derive from one another in an
industry can influence the value of their products (Podolny, 2005; Benjamin
and Podolny, 1999). As mentioned earlier, however, this line of work generally regards relational status as a signal of quality. It remains, in other
words, within the bounds of the quality certification framework. As a consequence, I here argue, it unnecessarily narrows our sociological understanding
of status. Research on status as a “prism” of the market (Podolny, 2001) essentially implements a classic definition of status as popularity among peers,
or deference received from them (e.g. Homans, 1951; Blau, 1964). And it
does so for good reasons: firms, after all, are probably better equipped than
9
customers to evaluate the quality of their peers, and one may legitimately
regard the deference they display to competitors as a signal of the latter’s
quality. Yet precisely because it frames status as a quality signal in the first
place, this approach also tends to obliterate another aspect of status, and
therefore fails to observe an important process of social valuation possibly
at play in markets.
Beside its dimension of popularity and deference indeed, status in the
social sciences has also long been viewed in terms of relational purity (e.g.
Marriott, 1959, 1968; DiMaggio, 1982). In this alternate view, status does
not emerge from one’s popularity with others, however weighted by the
social worth of one’s endorsers. Instead, it has its roots in the selectness of
one’s social affiliates, or said otherwise in the purity provided by one’s lack
of ties to status-threatening alters. This observation thus makes room for
two forms of status in markets, which for the sake of clarity one may refer
to as popularity and relational purity.
Do these two forms of status have independent effects on value in markets? Do they, in other words, underlie distinct social valuation processes?
This article shows this to be the case. It also argues that the notion of
relational purity, as applied to market intermediaries and the objects they
distribute, provides the analytical leverage to isolate a mechanism of social
valuation previously under-identified in the economic sociological literature
– namely consecration. It thus puts a name on the form of valuation associated with status as relational purity.
Consecration: theory and analytical strategy
Research on cultural valorization has repeatedly highlighted the importance
of consecration processes, whereby “some individuals or objects are collectively identified as worthy of veneration and esteem, whereas others are
not” (Allen and Parsons, 2006, p. 809; see also Bourdieu, 1993; Heinich,
1996; Corse and Griffin, 1997; Dowd et al., 2002; Allen and Lincoln, 2004;
English, 2005). Consecration, in this tradition, is usually seen as a selection process performed by an authoritative body, according to rigorous
procedures, and with an eye to identifying objective differences between the
individuals or objects that are selected and the ones that are not (Allen and
Parsons, 2006).5
This characterization may in effect suffice to empirically identify most
5
In their study of cultural consecration in baseball, for example, Allen and Parsons
illustrate how induction into the Baseball Hall of Fame obeys these various criteria.
10
consecration projects. In light of the two frameworks outlined earlier, however, one may wonder whether it does not too readily conflate consecration
with another social valuation process. If indeed consecration amounts to
the scrupulous selection, by a legitimate entity, of individuals or objects of
outstanding merit, it does not seem much different from quality certification by authoritative experts. Is consecration then just another word for
certification – or, for that matter, for the kind of social valorization associated with one’s endorsement by numerous others? Quite to the contrary,
I here argue that it captures a process analytically different from, though
oftentimes incidental to, these other valuation mechanisms.
To put things simply, the selection procedure through which some individuals or objects are identified as worthy of esteem while others are not
usually entails two closely intertwined processes. The most obvious is indeed a certification or endorsement process, applying to those individuals
or objects that are selected. Just as important, however, is the concomitant operation whereby selection draws a line between the individuals that
are selected and the ones are not. Selection therefore does not only alter
our perception of each single individual by touching them or not, and by
bestowing upon the chosen the authority of the selector. It also does so by
reconfiguring the whole set of individuals into a heterogeneous one, carving
out two exclusive and hierarchically ordered regions in the population – two
regions whose relation, in other words, is not unlike Durkheim’s vision of
the relation between the sacred and the profane (Durkheim, 1965). Through
their affiliation with one or the other of these regions, the relational identity of individuals is redefined: selected individuals are made similar to one
another and overtly different from non-selected ones, and vice versa. As far
as the impact of this redefinition on the value of individuals or objects is
concerned, I propose to refer to this second aspect of selection, and to it
only, as consecration.
Empirically, it may be hard to parse out what, in the impact of selection
on value, owes to certification and what to consecration proper. One reason – which may also explain their frequent confounding in the literature –
is that consecration often goes unnoticed, to the benefit of other valuation
mechanisms. Commenting on the selection process involved in rites of institution, Bourdieu thus remarks that “by solemnly marking the passage over
a line which establishes a solemn division in the social order, rites draw the
attention of the observer to the passage [that is, to the certification of the
individual’s quality], whereas the important thing is the line.” (Bourdieu,
1991, p. 118). More annoyingly, however, in many an instance the two processes occur jointly, and can simply not be disentangled in empirical data.
11
The selection of a writer to a prestigious prize, for example, may entail both
consecration – the writer now ranks among the prize’s laureates – and certification of the quality of her work. And it would be quite straightforward
to measure the specific impact of the award on, say, the value of her subsequent publishing contracts: one could for example compare these contracts
with those of counterparts of equal reputation or talent, but who failed to
earn the prize. Yet because the award simultaneously grants certification
and consecration, one would be hard-pressed to single out the contribution
of each in the emergence of a value difference between the laureate and her
peers.
If one is to properly identify the effects of consecration, one therefore
needs to separately capture an authority’s certification power and its propensity to consecrate. As far as the former is concerned, it does not seem unconceivable to approach it through certain observable attributes, such as
the expertise and reputation of the selecting entity, or the average quality of
the individuals it selects. The challenge, then, lies in the identification of a
distinct measure of consecration power – that is, of one’s ability to establish
a clear line between the individuals one selects and the ones one does not.
To devise that measure, I suggest we take advantage of the fact that
empirical settings often display more than one selecting authority, and that
individuals need not be selected by one of them only. When this is the case,
the relational purity of selecting entities effectively captures their consecration power.
Consider for example the selection of artists by galleries in the market
for contemporary art – an empirical case I will return to shortly. From an
analytical standpoint, consecration here occurs through the line an authoritative gallery draws between the individuals it decides to represent and the
individuals it does not. In practice, however, some of the artists chosen by
gallery a may also deal with gallery b, thereby weaving cross-cutting ties
between the two houses, and symbolically connecting the artists in and outside a. It thus becomes apparent how two equally authoritative galleries can
consecrate to a greater or lesser extent: this happens when the boundary
between the artists one gallery respectively selects and rejects is blurred by
the presence of cross-cutting ties with another gallery. In Figure 1 below, for
example, a’s consecration power is greater than that of equally authoritative
a1 , whose artists are partly shared with b.
We may in fact go one step further, if we assume that most galleries
somehow have to share artists with others – if only because artists have
a career, hence a track record of other galleries by whom they have been
previously selected. For gallery a, then, choosing to represent certain artists
12
B
A1
A
Figure 1: Consecration power as boundary enforcement
also amounts to drawing a divide between the galleries b it accepts to share
artists with, and the galleries c with whom it does not. When deciding on
its portfolio of artists, in other words, a simultaneously reconfigures the set
of other galleries into a dual structure, and assumes a position on one side
of that structure (see the left panel in Figure 2).
This position we can equivalently describe in terms of consecration power
or relational purity. The divide a introduces between other galleries (and
between associated sets of artists) can indeed be more or less clear-cut. It is
blurred, in particular, when the galleries a coopts also coopt other galleries
whom a does not coopt. In the terms I have used earlier, a’s position then
carries less consecration power. Two galleries a and a1 , in other words, have
greater or lesser consecration power if they have greater or lesser relational
purity, or engage in relations with respectively pure and impure galleries
– that is, with galleries who themselves connect only to galleries a and a1
deem worthy enough to deal with, or to galleries outside the realm of a and
a1 (see respectively the left and right panels in Figure 2). Thus conceived,
relational purity essentially offers a measure of consecration power.
C
C
D
A
A1
B
B
Figure 2: Consecration power as boundary enforcement or relational purity
13
The remainder of this article builds on this property to show that consecration is indeed how brokerage impacts value in the art market. The definition of purity it uses, it should be noted, departs from the idea of purity
as the rejection of economic motives deemed impure, in contrast to merely
artistic ones for example Bourdieu (1993). Contrary to that approach, purity here does not rest on the conformity of one’s actions or decisions with
an external guiding principle – or the avoidance of another. Rather, and in
accordance with the notion that consecration simultaneously generates the
social positions it also affiliates individuals with, purity here is essentially
a self-referential dimension, capturing an authority’s ability to enforce the
symbolic boundaries it has itself contributed to define.
To illustrate the impact of consecration on the economic worth of artists,
I first show that art market intermediaries do indeed line up in a status
hierarchy ordered by degrees of relational purity. I then proceed to measure
the specific impact of their consecration power on the economic value of
artists.
Case and method
The empirical setting this article focuses on is the French market for modern
art in early twentieth-century Paris. The reasons for choosing this case are
both historical and methodological. Up until the Great Depression indeed,
Paris was not only a prominent center of artistic creativity. It was also the
core of the worldwide market for the work of living artists. Like no other
city, it attracted a dense crowd of talents. And perhaps nowhere else were
the careers of these artists as systematically processed by a vivid population
of galleries (e.g.Gee, 1981; see also Ring, 1931).
Two features in particular make the period and locale valuable for studying the role of art market intermediaries in shaping the value of artists. Due
to their concentration in a small number of neighborhoods, first, contemporary art dealers typically knew each other and each other’s distribution
decisions closely – an aspect that surfaces time and again in dealers’ memoirs and testimonies (e.g. Blot, 1934; Granoff, 1949; Loeb, 1946; Ring, 1931;
Vollard, 1936; Weill, 1933). They were thus prone to managing their portfolios of artists relative to one another, a process I have put at the core of
the idea of consecration.
Second, the period was also one of high distrust in the ability of academic
institutions to successfully sanction the quality of contemporary artists. The
growingly dysfunctional academic system and its yearly Salon, slowly glut-
14
ted with an oversupply of young talents, had ceased to play this role over
the second half of the nineteenth century (White and White, 1965; Green,
1987; Mainardi, 1993; Jensen, 1994; Galenson and Jensen, 2002). Other,
independent salons mostly served to launch the careers of new artists (Gee,
1981). And with few exceptions, museums were not given much credit when
it came to selecting and rewarding talented modern artists. As a consequence, an active population of galleries and critics gradually took over the
role of public institutions as brokers between artists and collectors, and as
authorities defining artistic quality.6
Could that population of galleries also have been a source of consecration
for artists? In line with the analytical strategy laid out above, I address this
question by asking whether the network of cross-representations of artists
in Parisian galleries displayed a structure akin to the ones sketched out in
Figure 2. If this was the case – that is, if in that network galleries clearly
lined up according to a principle of greater or lesser relational purity –
then the market can be deemed to have provided a breeding ground for
consecration processes. Identifying the respective position of galleries in
that structure should therefore also provide a measure of the consecration
power they wielded.
Rather than directly devising a measure of each gallery’s consecration
power, I thus adopt a two-step approach, first describing the general organization of the market in an effort to assess its overall propensity to sustain
consecration processes. This initial step is designed to address a shortcoming frequently associated with research on networks as prisms through
which third parties can gauge the status of market actors. For ties and affiliations to work as prisms indeed, they need to be fairly easily interpreted
by observers. Studies of networks as the prisms of markets, however, rarely
ascertain the interpretability of relational patterns to potential beholders.
In contrast, I here assess this overall interpretability by first fitting a blockmodel to the network of cross-represen-tations of artists by Parisian art
dealers. This essentially isolates sets of galleries similar to one another as
far as their ties to counterpart galleries are concerned. Most importantly, the
stochastic blockmodeling technique I implement also provides information
on the clarity of the underlying relational data (Nowicki and Snijders 2001;
see Appendix A). It thus indicates whether dealers (or sets of structurally
equivalent dealers) indeed arranged according to a principle of greater or
lesser relational purity.
The data I build on report representation of artists in Parisian galleries
6
On that matter, see the opinions collected in Charensol (1996 [1925]).
15
in season 1928-29, for which the documentation is most systematic. The
population of galleries is defined through a conventional criterion, though
it eventually includes most modern art dealers active in Paris at the time.
Specifically, a gallery is included if it exhibited at least one living painter,
either on permanent or temporary display, over the season of interest. There
were 120 such galleries. Two sources then permit reconstructing the lists of
artists exhibited on a permanent basis by these galleries.7 Over the season at
stake, 665 painters are thus regarded as represented permanently by at least
one of the 120 dealers – an average of about 16 artists per gallery. Socioeconomic data were further collected on both galleries (opening date; capital;
form of business operation; participation as expert to auction sales for modern painting at Paris auction house, Hôtel Drouot; documented purchases of
modern artists at Drouot; publishing activity; and geographic location) and
artists (socio-demographic characteristics; stylistic affiliation; and prices at
Drouot in 1928-29).8 This information provides guidance to analyze the
blockmodel below.
The blockmodeling procedure is described in Appendix A. Table 1 displays the image-matrix of the blockmodel best suited to the network of
cross-representations. It reports the average number of artists shared by
two galleries in the various positions of that blockmodel, thereby offering
a summary view of the cross-representation network. Thus, two dealers in
position 1 share on average 1.6 artists, and dealers in position 1 typically
share 1.4 artists with dealers in position 2. The few galleries excluded from
the analysis at earlier stages because they did not share enough (position 7)
or any (position 8) artists with other galleries (see Appendix A) are not
shown. Table 2 reports the means of various socioeconomic variables for
galleries in each of the eight positions.
The selected blockmodel suggests a clear-cut structure in the network of
cross-representations across galleries. Two aspects are specially worth noting. First, one needs to emphasize the centrality of position 6 galleries in
the cross-representation network (highlighted through dark-shaded cells in
Table 1). This centrality essentially arises from the high number of artists
these galleries claim to represent: 33.4 on average, as opposed to a more
7
See La Semaine à Paris, published weekly over the period, and Fage (1930, 132-151).
The two sources generally agree on the artists represented by various galleries. When this
is not the case, however, artists mentioned by one source only were added to the list of
artists represented by a gallery.
8
Systematic business data come from Registre du Commerce de la Seine, Archives
de Paris, série D33U3. For information regarding sales and prices at Drouot see Lang
(1918-1929) and Gee (1981).
16
17
1.6
1.4
0.1
0.1
0.1
0.9
Position 1
Position 2
Position 3
Position 4
Position 5
Position 6
Position 1
4.5
0.2
0.9
2.0
3.0
-
Position 2
5.1
0.4
1.2
3.3
-
-
Position 3
2.1
0.3
0.5
-
-
-
Position 4
0.6
0.2
-
-
-
-
Position 5
8.3
-
-
-
-
-
Position 6
Table 1: Average number of artists shared by two galleries
in the various positions of the selected blockmodel. Light-shaded cells
indicate the status hierarchy discussed in the text. Dark-shaded cells highlight
the strong ties of merchant dealers with dealers of any other position.
Positions 7 and 8 (see Appendix A) are not shown.
18
15
24
16
18
18
13
7
120
3
4
5
6
7
8
All
11
1
2
Number of
galleries in the
position
Position
1920
1927
1926
1923
1918
1922
1921
1919
1897
Average
opening date
543,600 (39)
200,000 (1)
290,000 (2)
325,000 (8)
210,300 (4)
63,300 (3)
813,800 (8)
344,000 (5)
1,070,000 (8)
Average capital and
(number) of
incorporated
businesses
16.2
5.8
6.1
33.4
8.9
12.5
18.5
20.9
10.6
Average number
of artists on
permanent
display
1876
1883
1885
1877
1881
1880
1884
1865
1844
Average date of
birth of artists in
a gallery
80.0
100
91.2
82.4
88.2
87.9
92.1
64.1
24.2
Average percent of
a gallery’s artists
alive in 1929
Table 2: Socioeconomic characteristics of galleries in the various positions of the selected blockmodel
4,185
114
3,202
4,553
1,424
4,880
3,273
6,305
8,237
Average price of a
gallery’s artists
(francs for 100 cm2
of oil on canvas)
manageable 13.1 in galleries of other positions.9 Their geographical location
and opening date do not clearly distinguish position 6 dealers, neither do
the socioeconomic or stylistic characteristics of the artists they distribute.
Another striking feature, however, tells them apart: they take intensive part
as buyers in Hôtel Drouot’s auction sales. Among the forty-five dealers documented to have bought modern work at Drouot over the 1920s, thirteen (or
29%) are thus sorted in position 6 – which only makes up for 15% of the 120
galleries. In all likelihood, then, position 6 galleries do not really represent
the 33.4 artists they claim to exhibit on a permanent basis. Rather, they
merely can have access to their work, either through the auction market or
through second-hand purchases. Their market role, in other words, departs
from that of entrepreneur-dealers concerned with the promotion of a small
number of carefully selected artists on the primary market. Various pieces
of evidence instead suggest seeing them as trader-dealers operating on the
secondary market (Moulin, 1987).
Putting position 6 aside, however, cross-representations across galleries
do indeed delineate a structure analyzable in terms of varying consecration
power. More specifically, the pattern of ties between galleries of positions 1
to 5 displays the formal features of a status hierarchy – highlighted through
light-shaded cells in Table 1 – ordered by degrees of relational purity. Position 1 dealers thus typically cross-represent artists with other dealers in
position 1 and with galleries in position 2, yet not with those in positions 3
to 5. This takes on particular significance if one also notes that position 1
dealers can typically be regarded as the most established. As indicated in
Table 2, they have generally been in business for a longer time, their capital
is higher, and the artists they represent are on average both older and more
expensive that those in the galleries of other positions. Information on their
location further suggests that position 1 galleries cluster in more prestigious
Parisian neighborhoods. The relational boundary between galleries of positions 1 and 2 on the one hand and positions 3 to 5 on the other can thus
be seen as introduced by galleries whose authority and prestige are highest.
This boundary is in turn enforced with varying strength by the dealers of
position 1 and those of position 2. The latter indeed share artists with galleries in positions 3 and 4, while the former do not. Meanwhile, the selection
decisions of position 2 galleries, whose socioeconomic characteristics signal
as fairly established, also introduce a second boundary, between galleries
of positions 3 and 4 on the one hand, with whom they share artists, and
9
Differences in means across positions reported in this section are all significant according to appropriate tests.
19
galleries of position 5, with whom they do not. Position 3 and 4 dealers,
finally, are the only ones to significantly cross-represent artists also featured
in position 5.
As far as their selection decisions were concerned, then, Parisian galleries in the 1920s did in fact delineate a status hierarchy based on degrees
of relational purity. The market as a whole was thus prone to sustaining
consecration processes. The method I have used to uncover it, in addition,
permits locating each gallery in that purity hierarchy. It thereby provides a
hand on their consecration power. As compared to galleries’ ability to certify and define quality, did that power indeed influence the prices of artists?
And does that help us better understand what market intermediaries do to
the products they distribute?
The price of purity
Data
Auction prices of artists as an indicator of value
To answer these questions, I here turn to prices of artists at Paris most
important auction house, Hôtel Drouot, in the late 1920s. Such prices may
not be in total accordance with prices charged by galleries, yet they are
probably a fair indicator of what I am after here, namely the economic
value of an artist as determined by the supply of and demand for her works.
Various pieces of evidence thus suggest that there existed a strong correlation between gallery and auction prices. In addition, modern art auction
sales had become established enough at Drouot by the end of the 1920s
that they probably offer a more reliable barometer of the economic value
of artists. Historical evidence supporting these arguments is developed at
greater length in Appendix B below.
I model the selling prices at Drouot, over the season 1928-29, of artists
represented permanently in Paris galleries at the same period. Prices are
compiled from the last volume of Lang (1918-1929), which records all art
sales held at Drouot over the period. In the following analyzes, sales of
paintings are the only ones considered. Since I am interested in artist- and
gallery-related rather than technique-related predictors of value, this focus
has no dramatic consequences on the analysis.10 I thus use information on
10
Paintings form the bulk of the works by modern artists sold at auctions throughout
the 1920s. They are also the ones for which Lang most systematically reports size, a most
valuable piece information when it comes to modeling variations in prices.
20
the sales of 1,196 paintings by 173 unique artists with permanent gallery
representation in 1928-29.
Auction prices are modeled using hierarchical models, which permit appropriately adjusting for the effects of predictors operating at various levels.
Sales of individual artworks are indeed nested at the level of artists. As
sales are observed at auctions, however, they cannot be considered to nest
at the level of galleries. Gallery-related characteristics are therefore passed
on to the artists these galleries represent. It should also be remembered
that artists can be featured by several galleries. When this is the case,
gallery-level variables are expressed at the artist level as scores – that is, as
the average of the values they take in the various galleries representing the
artist.
Finally, two separate series of models are fit below. The first series
does not adjust for the past prices of artists, in line with both Rengers and
Velthuis (2002) and Beckert and Rössel (2004). While this allows for a better
comparison with these authors’ findings, it also leaves open the possibility
that the observed effects of gallery characteristics may arise from a selection
scenario whereby dealers with higher reputation, for example, would merely
choose to represent artists on the basis of their higher past prices. To rule
out this scenario, I thus fit a second series of models, where past prices are
included as an additional predictor of 1928-29 prices.
Measuring consecration power
Since I am mostly concerned with the role of intermediaries in shaping the
economic value of artists, gallery characteristics are the chief object of interest – and among them consecration power will obviously receive special
scrutiny. The latter is here captured as a gallery’s position in the status hierarchy uncovered earlier. That position essentially measures the relational
purity a gallery claims for its artists through the shape of its relations with
others. Practically, gallery purity is coded as decreasing at a constant rate
as one moves down one position in the status order. Thus, galleries have a
relational purity of 5 in position 1, 4 in position 2, 3 in positions 3 and 4,
and 2 in position 5.
The purity of galleries in positions 7 and 8 cannot be directly assessed
on the same grounds, since these galleries lack any affiliation to the status
hierarchy as presented in Table 1. Position 7 dealers, however, can easily be
reintroduced if one slightly loosens the threshold used to approach relations
between galleries. When this threshold is set to one rather than two shared
artists, and a blockmodel is fit to relations between galleries, 12 out of the
21
13 galleries in position 7 cluster with some of the galleries making up position 5 in Table 1, to form an even lower step in the hierarchy. I therefore
give position 7 galleries a status score of just 1. Position 8 galleries are similarly attributed the lowest possible score, on the grounds that they occupy
a hardly identifiable position in the market hierarchy.
Finally, relational purity is not straightforwardly established for traderdealers (position 6 in Table 1). On the one hand indeed these dealers show
artists also represented in both the highest and the lowest steps of the status
hierarchy, thereby positioning themselves somewhere in the middle. Yet
on the other hand their specific role as trader-dealers mostly operating on
the secondary market makes them poor candidates to confer any kind of
consecration. To solve this indeterminacy, two specifications of their status
were implemented. In the first one, trader-dealers were attributed a middleof-the-range status of 3. In the second, their status was left undefined.
Findings did not differ sharply depending on the adopted specification. The
results reported below use an undefined status for position 6 galleries.
Galleries as certifiers and qualifiers
As discussed earlier, a variety of processes involving galleries can contribute
to shape the value of artists beyond consecration itself. To disentangle
their respective impacts, additional gallery characteristics were introduced
as predictors of prices, capturing in particular their role in certifying and
defining artistic quality.
Various indicators can measure a gallery’s reputation, conceived as its
perceived ability to certify the quality of artists. Longevity in business is
one, as too many ill-advised representation decisions can entail the demise
of a gallery. Whether the gallery’s manager acts as expert at auction sales
is another: that one’s expertise is sought after for this job clearly signals
one’s recognized ability to assess artistic quality. To find out whether these
variables were indeed correlated, and ultimately to build a synthetic index of
reputation, I conducted a principal component analysis of gallery-level data
(Figure 3). Opening date, incorporation (variable “inc” in Figure 3), and in
case a gallery was incorporated its capital in 1928-29, were collected from
the Registre du Commerce de la Seine in Paris departmental archives.11
Other variables come from miscellaneous sources: the number of times a
gallery’s manager acted as expert at Drouot auction sales in 1928-29 (variable “nbexpert”) was obtained from Lang (1918-1929). Whether a gallery
11
Registre du Commerce de la Seine, Archives de Paris, série D33U3.
22
1.0
had a periodical advertising organ, such as a bulletin or journal, or alternatively published monographs on its artists, is reported in variable “edit”.
Finally, the 1928 and 1929 issues of the Bottin mondain, a directory of Paris
highly regarded personalities and businesses, were searched for the presence
of galleries or dealers (variable “bottin”). Figure 3 displays the first factor
map for this set of variables.
0.0
bottin
nbexpert
inc
opening
capital
-0.5
Dim 2 (16.62%)
0.5
nbpainters
-1.0
edit
-1.0
-0.5
0.0
0.5
1.0
Dim 1 (36.96%)
Figure 3: Principal component analysis of
gallery-level variables: first factor map
As one can see, longevity in business (the opposite of opening date)
strongly correlates with the number of times a gallery’s manager acted as
expert at Drouot auctions. These two variables actually contribute the
most to the construction of dimension 1 in the factor map. This dimension,
therefore, may be regarded as capturing the reputation of dealers. That a
gallery’s presence in the Bottin mondain also correlates with dimension 1
further confirms this interpretation. When modeling the impact of galleries
on prices, I shall use their score along this dimension as an index of their
reputation. Interestingly, capital also strongly correlates with dimension 1
– too strongly indeed to be used as an additional variable in a regression
framework. Capital, in the models below, is therefore considered an element
of a gallery’s reputation.
That a gallery regularly published a journal or monographs – or both –
was finally used as a proxy for the promotion it reserved to its artists,
and thus for its ability to shape collectors’ preferences through qualification
work. As shown in Figure 3 (variable “edit”, the largest contributor to
dimension 2), this aspect of the activity of galleries is hardly associated with
23
their reputation. Yet it negatively correlates with the number of artists they
represent (“nbpainters”). This dovetails with the notion that publishing
captures promotional efforts: the larger the number of artists a dealer has
to handle, the less promotion each of them should logically receive.
Artists’ and artworks’ characteristics
While characteristics of artists and artworks can be regarded as controls used
to properly disentangle the role of intermediaries, they are also interesting
in themselves. Since all works in the sample were paintings, technique does
not intervene here at the work level. The main variable of interest at this
level is therefore size. Price is expected to rise with the size of paintings (e.g.
Sagot-Duvauroux et al., 1992), yet the strength of this relationship may also
differ across artists. For successful artists in particular, whose mere name
makes for a larger part of a painting’s price, size should not matter relatively
as much. Allowing the slope of the relationship between size and price to
vary across painters might therefore help refine the modeling of prices.
Artists’ demographics are also expected to impact the selling price of
their production. Age (or more generally the number of years elapsed since
an artist’s birth) is thus expected to drive prices up, since it is both a signal of
quality – artists represented by galleries at later ages have passed a tougher
test of time (Hume, 1757) and still managed not to fall into oblivion – and a
proxy for the audience of collectors an artist might have been able to build
over time (Bowness, 1990). In addition, the death of an artist, which creates
an immediate shortage of her available work, should positively influence her
selling prices, even when date of birth is adjusted for.12
The number of works of an artist auctioned over the season is further
expected to have a depressing effect on prices – in a basic instantiation of the
law of supply and demand. One can also anticipate the number of galleries
representing an artist to influence prices – albeit not necessarily in a linear
way. Being shown by numerous dealers should indeed push prices upward:
directly, it increases the likelihood that one’s dealers will show up at auctions
to support prices, if need be. Indirectly, it may also bolster the demand
for one’s work, since various dealers do not necessarily reach out to the
12
Gender was also included as a predictor, but never found to have a significant effect on
prices. This should not however be seen as evidence that being female did not constitute
an important liability in the art market. Rather, the absence of significant impact arises
from a lack of observations, which in turn points to the fact that very few women made it
into the population of artists enjoying gallery representation and regular trading of their
work at Drouot at the time. Of the 173 artists with both representation and trading
in 1928-29 analyzed below, only 13 were women.
24
same collectors. If too many galleries represent an artist, however, collusion
between them might prove tough to maintain, and price competition, which
is usually reined in on the primary market, might resurface: dealers unable
to sell an artists’ output because it is already available from too many others,
will likely engage in undercutting competitors. While this should first drive
value down on the primary market, it could also have indirect consequences
on prices at auction.
Finally, one needs to model the influence of one more variable operating
at the artist level – namely the overall perceived quality of an artist’s work,
as signaled by the assessments of contemporary critics. To approach this
dimension, I here rely on a unique resource. In the fall of 1925, art journal
L’Art vivant organized a survey in which 64 art critics, with a wide range
of aesthetic inclinations, were asked to name the 10 artists they considered
most worthy of entering a projected museum of modern art – since no such
museum existed in Paris at the time. The survey results were presented in
Charensol (1996 [1925]), and are reported here in Appendix C. While the
survey was held in season 1925-26, I assume that the results would not have
been greatly different three seasons later. I therefore use the number of votes
received by each artist in the survey (variable “votes Art vivant” below) as
a measure of their critical recognition. In the market for modern art, critics
would typically review exhibitions for newspapers or art magazines. They
could also provide forewords to the catalogs galleries issued when showing
an artist or a group – and good reviews, or prefaces by important critics,
were certainly helpful in getting noticed by potential buyers. As unbiased
judges of the quality of artists, however, even the most prominent critics,
taken individually, were not necessarily regarded as worthy of unlimited
trust. Indeed, as Turpin observed, important conflicts of interest could arise
between their role as prescribers and their personal ties to various artists
(Turpin, 1929). Using a survey of a wide panel of critics by contrast makes
it possible to correct for individual biases.
Results
Descriptive statistics and correlations for all artist-level predictors, including
gallery-related characteristics passed on to artists as scores, are displayed in
Table 3 for the 173 artists with auction sales in 1928-29.13 Table 4 reports
13
It is worth noting here that the artists in the analysis are not representative of the overall population of artists with permanent gallery representation in 1928-29. The 173 artists
with auction sales in 1928-29 were on average older (47.5 years old vs. 44), were represented by more galleries (6.2 vs. 2.8) and enjoyed better critical acclaim (1.9 votes vs.
25
results as a series of models predicting the logged prices of artworks at
auctions that year. Model 1 is an empty model, and models 2 to 4, each
including additional predictors, gradually improve the overall fit.14 Model 5
allows the slope of the size predictor to vary across artists. Quite logically,
augmenting the models with artworks’ characteristics reduces unexplained
variance at the artwork level, and adding artists’ characteristics – or galleryrelated characteristics expressed at the artist level – helps account for artistlevel variance. Altogether, model 4, the most powerful among the first four
models, accounts for around 60% of the initial variance in the data, as
captured by empty model 1. This full model, in addition, fares better in
explaining variance at the artist level (70% of the initial variance explained)
than at the artwork level (26%). This likely arises from the relatively small
number of predictors introduced at the artwork level.
Effects of predictors at the work and artist levels are all significant and
in the expected direction, and their magnitude does not differ sharply across
models. A painting’s size positively affects its price, although this relationship loosens for artists whose prices are generally higher, as shown by the
negative correlation between intercept and size coefficient in model 5. This
is in accordance with the notion that successful artists charge a larger part
of their prices for their mere name, which a work bears regardless of its size.
The age and death of artists also drive prices up, while having more works
sold at auction over the season depresses them. Also in line with expectations, representation by a larger number of galleries yields an increase in
selling prices, but this positive relationship fades as the number of representing galleries gets larger: the coefficient associated with the squared number
of representing galleries is negative. Finally, critical acclaim is positively
correlated with price: each extra vote received in the survey by L’Art vivant
corresponds to a 13% (e0.12 = 1.13) increase in an artist’s auction prices.
Most important to the purpose of this article, however, is the impact
on prices of gallery-related characteristics. This is where one gets the traction to decide whether intermediaries indeed influence the economic worth
0.6). They were also featured by galleries with significantly higher reputation, though
not with larger advertising resources or higher consecration power. Despite this lack of
representativeness, I interpret statistical significance below, because not all artists with
similar characteristics as these 173 had works sold at auction in 1928-29.
14
Mean deviance and dic, two measures of fit suited to the evaluation of multilevel
models, are displayed at the bottom of Table 4. While they differ slightly – mean deviance
is a pure measure of fit, dic assesses a model’s predictive power by correcting for its
effective number of parameters – both indicate better fit when taking lower values. Also
shown is the variance left unexplained by various models, broken down by level (artworks
and artists).
26
27
1. Years elapsed since date of birth
2. Dead
3. Number of works sold in 1928-29
4. Number of galleries representing
5. Number of galleries representing (squared)
6. Votes Art vivant
7. Journal or publisher score
8. Reputation index score
9. Consecration power score
Mean
SD
Min
Max
1
1
0.436
0.060
0.080
0.056
0.192
0.014
0.329
0.381
47.61
11.90
22
80
1
-0.048
-0.052
-0.043
-0.107
-0.073
0.160
0.188
0.10
0.31
0
1
2
1
0.482
0.468
0.227
-0.101
-0.036
0.113
6.91
9.33
1
89
3
1
0.930
0.597
-0.064
-0.071
0.109
6.24
7.46
1
44
4
1
0.551
-0.040
-0.047
0.057
94.35
251.73
1
1,936
5
1
0.118
0.081
0.119
1.87
4.21
0
26
6
1
0.287
-0.025
0.08
0.18
0.00
1.00
7
8
1
0.439
0.45
2.32
-1.40
14.00
Table 3: Correlations and descriptive statistics for variables in the analysis (N = 173)
1
3.04
0.94
1.00
5.00
9
Table 4: Multilevel models predicting prices (logged)
Predictors
1
2
Model
3
Characteristics of artworks
Size (log)
–
Characteristics of artists
Years elapsed since date of birth
–
–
Dead
–
–
Number of works sold in 1928-29
–
–
Number of galleries representing
–
–
Number of galleries representing (sq.)
–
–
Votes Art vivant
–
–
Characteristics of galleries
Journal or publisher
–
–
–
Reputation index
–
–
–
Consecration power
–
–
–
Intercept
.59**
(.03)
5
.60**
(.03)
.60**
(.03)
.62**
(.05)
.03*
(.01)
.57*
(.25)
-.017*
(.008)
.19**
(.02)
-.004**
(.001)
.13**
(.02)
.02*
(.01)
.54*
(.24)
-.018*
(.007)
.20**
(.02)
-.004**
(.001)
.12**
(.02)
.02**
(.01)
.56**
(.24)
-.019*
(.007)
.20**
(.02)
-.004**
(.001)
.12**
(.02)
.16
(.38)
.07*
(.03)
.20**
(.08)
.17
(.38)
.07*
(.03)
.20**
(.08)
2.34**
(.26)
2.10**
(.37)
7.19** 2.47** 2.34**
(.11)
(.28)
(.26)
Correlation Intercept / Slope of log(Size)
4
–
–
–
–
-.95
DIC
Mean deviance (-2 loglikelihood)
3,119
3,122
2,825
2,833
2,604
2,657
2,576
2,638
2,550
2,611
Total variance
Variance artworks
Variance artists
N Works
N Artists
2.41
0.54
1.87
1,196
173
2.34
0.40
1.94
1,196
173
1.03
0.40
0.63
1,196
173
0.96
0.40
0.56
1,196
173
6.08
0.37
5.62
1,196
173
** p <.01; * p <.05; standard errors in parentheses.
28
of their products – and if so, how. Models 4 and 5 thus include the three
variables designed to test the respective effects of intermediaries’ certification, qualification, and consecration power. As measured through galleries’
promotional resources, qualification does not seem to have a decisive impact
on prices: although the coefficient is again in the expected direction – among
the artists under consideration, one’s work typically makes around 18% more
at auction if one is represented by galleries all having advertising organs, as
opposed to galleries all lacking them – statistical significance is poor. In
contrast, the reputation of art market intermediaries, capturing their ability
to certify the quality of their artists, is clearly associated with higher prices.
So is also their consecration power: even when adjusting for artist-related
characteristics and for the reputation and promotional activity of galleries,
a one-unit increase in one galleries’ consecration power – that is, for an
artist, being represented by galleries one step higher in the relational status
hierarchy uncovered above – increases prices by about 22%.
As far as the influence of intermediaries is concerned, however, one may
deem this first series of models unsatisfying. The results reported in Table 4
might indeed stem from a different scenario than the one where galleries
causally influence artists’ prices. It could in particular be the case that
galleries with higher reputation or consecration power merely pick, among
artists with similar characteristics, those whose prices are higher in the first
place. The observed association between reputation or consecration power
and prices would then arise from upstream selection, not downstream influence.
To rule out this selection scenario, I fit a second series of models, featuring past prices of artists as an additional predictor of 1928-29 prices. I thus
estimate the net impact, among artists with comparable attributes including previous commercial success, of either being represented one more year
by a gallery with certain characteristics, or being picked by one that year.
Past prices are measured as the average price per square centimeter sold by
an artist at Drouot in 1927-28. This has the downside of sharply decreasing
the number of observations, to 796 works by 93 artists whose paintings were
sold both in 1927-28 and 1928-29. Table 5 reports descriptive statistics and
correlations across predictors for these 93 artists.15
15
Incidentally, all of these artists were alive in 1929, making death an irrelevant predictor
here. They were also slightly older on average than the initial 173 (48.5 years old vs. 47.5),
garnered more critical acclaim (3 vs. 2 votes), were represented in more galleries (8 vs. 6)
and sold more works at auctions in 1928-29 (9 vs. 6). However, none of the differences in
means between the two groups is statistically significant at the p <.05 level. Accordingly,
the conclusions of the second series of models can be regarded as extensible to the larger
29
The second series of models is presented in Table 6. Not surprisingly, the
new explanatory variable, past prices, is highly predictive of 1928-29 prices.
The magnitude of the relationship also makes sense: a one franc difference
in price per square centimeter in 1927-28 yields a 80% difference in 192829 prices (e0.59 = 1.80). With the average price per square centimeter for
the 93 artists around 1.3 franc in 1928-29, this means a difference in the
range of one franc that year: artworks one franc higher in 1927-28 were thus
also roughly one franc higher in 1928-29.
Coefficients of gallery-related predictors can now more safely be interpreted as the causal influence on value of the representation of artists by
market intermediaries. Models 6 and 7 here show that galleries with higher
promotional resources do not have any noticeable impact on prices. The
coefficient is statistically insignificant, and its magnitude is virtually zero.
The same is true of galleries with higher reputation – our measure of
certification power. For artists previously commanding similar prices, there
is no clear-cut gain to being represented one more year by a more reputable
gallery, or picked by one that year. The positive relationship observed earlier between prices and galleries’ reputation can therefore be attributed to
a selection mechanism, whereby more reputable galleries tend to feature expensive artists. This dovetails with the findings by Beckert and Rössel on
the one hand, and Rengers and Velthuis on the other, that the certification
power of galleries has no noticeable impact on prices.
Consecration power, in contrast, retains most of its impact in models 6
and 7. Among artists with comparable characteristics – including past
prices – being represented one more year by a gallery one step higher in
the hierarchy based on relational purity, or being selected by one that year,
increases prices by about 20% (e0.18 = 1.20). In the absence of observations
on representation before 1928-29, one cannot disentangle what in that rise
owes to fresh selections by higher status galleries in 1928-29, and what to the
continuing effect of earlier selections. In any event, though, the consecration
power of galleries does have a clear, large, and positive impact on the prices
of their artists.
Consecration power also fares better to explain prices than the status
of market intermediaries classically approached through their popularity in
a deference network. This is the sense of models 8 and 9, which feature
measures of status both as popularity and as relational purity. Following
in the steps of Podolny (2001), popularity status is here measured as the
centrality of a gallery in the network of affiliations across dealers, weighted by
group of 173 artists.
30
31
1. Years elapsed since date of birth
2. Number of works sold in 1928-29
3. Number of galleries representing
4. Number of galleries representing (squared)
5. Votes Art vivant
6. Journal or publisher score
7. Reputation index score
8. Consecration power score
9. Bonacich’s centrality score
10. Prices 1927-28
Mean
SD
Min
Max
1
1
0.022
0.076
0.050
0.294
0.119
0.349
0.342
0.024
0.348
48.54
10.09
27
77
1
0.515
0.510
0.236
-0.139
-0.081
0.139
0.187
-0.085
8.559
10.94
1
89
2
1
0.938
0.590
-0.143
-0.149
0.090
0.346
0.350
8.462
9.08
1
44
3
1
0.549
-0.093
-0.104
0.046
0.219
0.248
153
325
1
1,936
4
1
0.114
0.053
0.135
0.235
0.819
3
5.11
0
26
5
1
0.299
0.043
0.009
0.036
0.10
0.20
0.00
1.00
6
Table 5: Correlations and descriptive statistics (N = 93)
1
0.472
-0.041
0.064
0.66
2.39
-1.30
14.00
7
1
0.552
0.197
3.09
0.90
1.00
5.00
8
1
0.172
11.89
4.59
0.00
17.53
9
1
1.45
2.73
0.04
16.00
10
Table 6: Multilevel models predicting prices (logged),
including past prices as a predictor
Model
Predictors
6
Characteristics of artworks
Size (log)
7
8
9
.60**
(.04)
.64**
(.05)
.60**
(.04)
.014*
(.0073)
-.006
(.006)
.09**
(.02)
-.0018**
(.0006)
.046*
(.018)
.58**
(.08)
.016
(.0074)
-.006
(.006)
.09**
(.02)
-.0018**
(.0006)
.044*
(.019)
.60**
(.08)
.0139
(.0075)
-.006
(.006)
.10**
(.03)
-.0018**
(.0006)
.047*
(.019)
.57**
(.08)
.0149*
(.0075)
-.006
(.006)
.10**
(.03)
-.0018**
(.0006)
.044*
(.019)
.59**
(.08)
.02
(.32)
.055
(.033)
.18*
(.08)
–
-.02
(.33)
.047
(.034)
.18*
(.08)
–
.05
(.33)
.049
(.035)
.22*
(.11)
-.01
(.02)
-.01
(.33)
.040
(.036)
.24*
(.11)
-.01
(.02)
1.06
(.55)
.73
(.63)
1.06
(.55)
.72
(.64)
–
-.98
–
-.98
DIC
Mean deviance (-2 loglikelihood)
1,539
1,605
1,516
1,581
1,533
1,605
1,509
1,580
Total variance
Variance artworks
Variance artists
N Works
N Artists
0.61
0.37
0.24
796
93
5.96
0.34
5.54
796
93
0.61
0.37
0.24
796
93
6.12
0.34
5.69
796
93
Characteristics of artists
Years elapsed since date of birth
Number of works sold in 1928-29
Number of galleries representing
Number of galleries representing (sq.)
Votes Art vivant
Price 1927-28
Characteristics of galleries
Journal or publisher
Reputation index
Consecration power
Bonacich’s centrality
Intercept
Correlation Intercept / Slope of log(Size)
** p <.01; * p <.05; standard errors in parentheses.
32
.64**
(.05)
the centrality of its affiliates (Bonacich, 1987).16 It essentially captures the
quality of a gallery’s artists, as signaled through the willingness of its peers
to also represent them. As can be seen from models 8 and 9, this signaled
quality does not significantly influence prices.17 Status as relational purity
and consecration power, in contrast, retains all of its explanatory power in
the two models, strengthening the finding that consecration is indeed how
galleries generate value for their artists.
Finally, and although they do not directly pertain to the main argument
of this research, it is worth commenting on the influence of artist-related
predictors in this second series of models. Representation by one additional
dealer in 1928-29 thus continues to drive prices up when past prices are adjusted for. This effect now clearly captures what a more widespread gallery
representation actually does to the value of artists, either through increased
support at the auction house, or through the construction of a larger audience for artists. Among artists with similar prices in 1927-28, older ones also
command higher prices at auctions in 1928-29. This is an indication that
prices grow with time at a faster pace for older artists than for their younger
counterparts – possibly because the mechanisms involved in the causal effect of age are not linear, for example if audiences build over time through
a snowballing process, or because new mechanisms, such as the anticipation
of death, intervene as age increases. Likewise, critical acclaim also seems to
work on the rate of growth of artists’ prices, so that among painters with
similar prices in 1927-28, those whose expert evaluation was better in 1925
enjoy even dearer prices in 1928-29.
Conclusion and implications
The market for modern art in early twentieth-century Paris offers a startling
illustration of consecration as a social process of valuation. For Matisse, Pi16
Specifically, status as popularity is here calculated as:
S(α, β) =
∞
X
αβ κ W κ+1 1,
κ=0
where α is a scaling coefficient, β is a weighting parameter, W is the matrix recording
affiliation ties between galleries, and 1 is a column vector where each element has the
value “1”. S(α, β), then, is also a column vector, the elements of which record Bonacich’s
centrality for each gallery in the affiliation network.
17
This is not the case either in models where relational purity is not included among
gallery-related variables affecting prices. Such models are available from the author upon
request.
33
casso and their peers, art market intermediaries did not merely facilitate the
encounter of their artistic output with the demand of collectors. They also
actively contributed to the generation of their value. They did so, however,
in a way that has generally been under-theorized by economic sociology.
Most of our knowledge on social mechanisms of valuation is indeed concerned with two broad types of processes: certification, or the signaling of
underlying quality, and qualification, or the social construction and diffusion of criteria for value. Neither of those processes was involved in the
production of value by intermediaries in the market for modern art. Their
certification power, however approached, did not noticeably influence art
prices, and neither did their qualification work. In contrast, consecration
power, conceived as the ability of more prestigious dealers to establish clear
lines between their artists and others, significantly enhanced the value of
artists.
This article proposed a theory of consecration as the affiliation of individuals to relationally pure strata in a structure made heterogeneous by the
selection decisions of intermediaries. In doing so, it insisted on the structural dimension of consecration. Taking the notion seriously, it revived the
Durkheimian insight that the sacred, religious or otherwise, should essentially be defined through its heterogeneity with the profane. In the market
for modern art, consecration thus ultimately arose from the structural divides introduced by intermediaries between various sets of artists.
This structural approach helped delineate consecration as a specific process of valuation in markets. Empirically, this enabled me to solve a puzzle
in economic sociology, by thinking anew about how intermediaries shape
the value of the products they distribute. From a more theoretical perspective, it highlighted the difference between consecration proper and the
certification mechanism associated with one’s selection by a legitimate or
reputed entity. Consecration here departs from the mere signaling of valued
properties concurrently entailed by selection. It instead operates through
the relational display of differences and identities between various sets of
objects or actors.
But that concept of consecration also holds insights for advancing our understanding of social valuation more generally (Lamont, 2012). As a matter
of fact, it recasts the way we often articulate cultural and structural arguments to account for social status and inequality. Approaches to the cultural
origins of social hierarchies classically view the latter as partly rooted in the
symbolic scales and boundaries actors routinely deploy to rank and categorize objects, people or groups. Under certain circumstances, such classification processes can result in unequal access to resources across social entities,
34
in the emergence of actual structural boundaries between them, and ultimately in social inequality (Lamont, 1992; Zuckerman, 1999; Tilly, 1998; Lamont and Molnár, 2002; Velthuis, 2005; Pachucki et al., 2007). The concept
of consecration I introduced, while not incompatible with this view, illuminates a reverse mechanism for inequality formation. Structural boundaries,
it suggests, do not only result from the application of preexisting metrics
or categories to the elements of a social system. Conversely, the structural
divides observed in a social setting – such as the ones market intermediaries
introduced between artists when making distribution decisions – can also
shape the overall perception of that setting’s hierarchy. They can, in other
words, concatenate into a valuation scale of their own, thereby acquiring a
distinct influence on the worth of various social entities. Thus, the relational
purity conferred by galleries – a structural form of purity, as opposed to the
kind granted by one’s compliance with culturally defined purity principles
(e.g. Dumont, 1970; Abbott, 1981) – was here found to bear specifically on
the value of artists.
At its core, then, the account of consecration developed in this study
points to the propensity of social structure itself to act as an independent
source of identity for the elements of a social system. This has implications
for research at the crossroad of cultural and structural social science that go
beyond issues of social valuation and inequality. That observed structural
patterns can provide a basis for the cultural constructs we mobilize when interpreting reality already lay at the heart of Durkheim’s account of cognitive
categories (Durkheim, 1965; Durkheim and Mauss, 1963), and the idea that
our minds internalize external distinctions in the social world is for example pivotal to Bourdieu’s theory of practice (Bourdieu, 1977, 1984). Recent
research on the interplay of social and symbolic structures, however, has
yet to produce an empirically grounded illustration of that idea (Pachucki
and Breiger, 2010). The present article has taken one step in precisely that
direction, showing how an observable system of divisions and groupings in
the social world can reverberate in the way its elements are apprehended,
appreciated, and ultimately valued.
There are finally limitations to what this article could achieve in that
domain. In particular, while the blockmodeling strategy I deployed made
it possible to ascertain the clarity of structural patterns observable in the
social world, it did not directly evidence the translation of such patterns
into mental structures available for the interpretation and appreciation of
social reality. Instead, I had to implicitly posit that translation. If future
research on the interaction of social and symbolic structures is to rest on
firmer ground, it should strive to test that assumption more systematically.
35
Only then shall we arrive at an empirically sound vindication of old ideas on
structural patterns, symbolic structures, and their intricate relationships.
36
References
Abbott, Andrew. 1981. “Status and Status Strain in the Professions.” American Journal of Sociology 86:819–835.
Adler, Moshe. 1985. “Stardom and Talent.” American Economic Review
75:208–212.
Allen, Michael Patrick and Anne E. Lincoln. 2004. “Critical Discourse and
the Cultural Consecration of American Films.” Social Forces 82:871–894.
Allen, Michael Patrick and Nicholas L. Parsons. 2006. “The Institutionalization of Fame: Achievement, Recognition, and Cultural Consecration in
Baseball.” American Sociological Review 71:808–825.
Anteby, Michel. 2010. “Markets, Morals, and Practices of Trade: Jurisdictional Disputes in the U.S. Commerce in Cadavers.” Administrative
Science Quarterly 55:606–638.
Basler, Adolphe. 1926. La peinture: religion nouvelle. Paris: Librairie de
France.
Becker, Howard S. 1982. Art Worlds. Berkeley: University of California
Press.
Beckert, Jens and Patrik Aspers (eds.). 2011. The Worth of Goods: Valuation and Pricing in the Economy. New York: Oxford University Press.
Beckert, Jens and Jörg Rössel. 2004. “Kunst und Preise: Reputation als
Mechanismus der Reduktion von Ungewissheit am Kunstmarkt.” Kölner
Zeitschrift für Soziologie und Sozialpsychologie 56:32–50.
Benjamin, Beth and Joel Podolny. 1999. “Status, Quality and Social Order in the California Wine Industry.” Administrative Science Quarterly
44:563–589.
Bielby, William T. and Denise D. Bielby. 1999. “Organizational Mediation of Project-Based Labor Markets: Talent Agencies and the Careers of
Screenwriters.” American Sociological Review 64:64–85.
Blau, Peter. 1964. Exchange and Power in Social Life. New York: John
Wiley & Sons.
Blot, Eugène. 1934. Histoire d’une collection de tableaux modernes : cinquante ans de peinture (1882 - 1932). Paris: Editions d’art.
37
Bonacich, Phillip. 1987. “Power and Centrality: A Family of Measures.”
American Journal of Sociology 92:1170–1182.
Bonus, Holger and Dieter Ronte. 1997. “Credibility and Economic Value in
the Visual Arts.” Journal of Cultural Economics 21:103–118.
Bourdieu, Pierre. 1977. Outline of a Theory of Practice. Cambridge: Cambridge University Press.
Bourdieu, Pierre. 1984. Distinction: A Social Critique of the Judgment of
Tastes. Cambridge: Harvard University Press.
Bourdieu, Pierre. 1991. Language and Symbolic Power. Cambridge: Harvard
University Press.
Bourdieu, Pierre. 1993. The Field of Cultural Production.
Columbia University Press.
New York:
Bowness, Alan. 1990. The Conditions of Success: How the Modern Artist
Rises to Fame. London: Thames and Hudson.
Breiger, Ronald. 1974. “The Duality of Persons and Groups.” Social Forces
53:181–190.
Breiger, Ronald, Scott A. Boorman, and Phipps Arabie. 1975. “An Algorithm for Clustering Relational Data with Applications to Social Network Analysis and Comparison with Multidimensional Scaling.” Journal
of Mathematical Psychology 12:328–383.
Charensol, Georges. 1996 [1925]. Pour un musée français d’art moderne. Une
enquête de L’Art vivant en 1925 . Paris: Réunion des Musées Nationaux
& Séguier.
Corse, Sarah M. and Monica D. Griffin. 1997. “Cultural Valorization and
African-American Literary History: Re-Constructing the Canon.” Sociological Forum 12:173–203.
Dewey, John. 1939. Theory of Valuation. Chicago: The University of
Chicago Press.
DiMaggio, Paul. 1982. “Cultural Entrepreneurship in Nineteenth-Century
Boston, Part II: The Classifiation and Framing of American Arts.” Media,
Culture and Society 4:303–322.
38
DiMaggio, Paul and Hugh Louch. 1998. “Socially Embedded Consumer
Transactions: For What Kinds of Purchases Do People Most Often Use
Networks?” American Sociological Review 63:619–637.
Doreian, Patrick, Vladimir Batagelj, and Anuska Ferligoj. 2005. Generalized
Blockmodeling. Cambridge: Cambridge University Press.
Douglas, Mary. 1966. Purity and Danger: An Analysis of Concepts of Pollution and Taboo. London: Routledge and Kegan Paul.
Dowd, Timothy J., Kathleen Liddle, Kim Lupo, and Anne Borden. 2002.
“Organizing the Musical Canon: The Repertoires of Major U.S. Symphonie Orchestras, 1842 to 1969.” Poetics 30:35–61.
Dumont, Louis. 1970. Homo Hierarchicus: An Essay on the Caste System.
Chicago: The University of Chicago Press.
Durkheim, Emile. 1965. The Elementary Forms of Religious Life. New York:
The Free Press.
Durkheim, Emile and Marcel Mauss. 1963.
Chicago: The University of Chicago Press.
Primitive Classification.
English, James F. 2005. The Economy of Prestige: Prizes, Awards, and the
Circulation of Cultural Value. Cambridge: Harvard University Press.
Espeland, Wendy N. and Michael Sauder. 2007. “Rankings and Reactivity: How Public Measures Recreate Social Worlds.” American Journal of
Sociology 113:1–40.
Espeland, Wendy N. and Mitchel L. Stevens. 1998. “Commensuration as a
Social Process.” Annual Review of Sociology 24:313–343.
Fage, André. 1930. Le collectionneur de peintures modernes. Comment
acheter, comment vendre. Paris: Les Editions Pittoresques.
Fourcade, Marion. 2011. “Cents and Sensibility: Economic Valuation and
the Nature of ‘Nature’.” American Journal of Sociology 116:1721–1777.
Galenson, David W. and Robert Jensen. 2002. “Careers and Canvases:
The Rise of the Market for Modern Art in Nineteenth Century.” NBER
Working Paper 9123.
39
Gee, Malcolm. 1981. Dealers, Critics and Collectors of Modern Painting.
Aspects of the Parisian Art Market between 1910 and 1930 . New York:
Garland Publishing.
Geertz, Clifford. 1978. “The Bazaar Economy: Information and Search in
Peasant Marketing.” American Economic Review 68, Supplement:28–32.
Granoff, Katia. 1949. Histoire d’une galerie. Paris: Galerie Katia Granoff.
Green, Nicholas. 1987. “Dealing in Temperaments: Economic Transformation of the Artistic Field in France During the Second Half of the
Nineteenth Century.” Art History 10:59–78.
Hatchuel, Armand. 1995. “Les marchés à prescripteurs. Crise de l’échange
et genèse sociale.” In L’inscription sociale du marché, edited by Annie
Jacob and Hélène Vérin, pp. 205–225, Paris. L’Harmattan.
Heinich, Nathalie. 1996. The Glory of Van Gogh: An Anthropology of Admiration. Princeton: Princeton University Press.
Homans, George C. 1951. The Human Group. London: Routledge & Kegan
Paul.
Hsieh, Mo-Han and Christopher L. Magee. 2008. “An Algorithm and Metric for Network Decomposition from Similarity Matrices: Application to
Positional Analysis.” Social Networks 30:146–158.
Hume, David. 1757. Four Dissertations. I. The Natural History of Religion.
II. Of the Passions. III. Of Tragedy. IV. Of the Standard of Taste. London:
A. Millar.
Hutter, Michael, Christian Knebel, Gunnar Pietzner, and Maren Schäfer.
2007. “Two Games in Town: A Comparison of Dealer and Auction Prices
in Contemporary Visual Arts Markets.” Journal of Cultural Economics
31:247–261.
Jensen, Robert. 1994. Marketing Modernism in Fin-de-Siècle Europe.
Princeton: Princeton University Press.
Karpik, Lucien. 2010. Valuing the Unique: The Economics of Singularities.
Princeton: Princeton University Press.
Khaire, Mukti and R. Daniel Wadhwani. 2009. “Changing Landscapes: The
Construction of Meaning and Value in a New Market Category – Modern
Indian Art.” Working paper, Harvard Business School.
40
Lamont, Michèle. 1992. Money, Morals, and Manners: The Culture of the
French and the American Upper-Middle Class. Chicago: The University
Of Chicago Press.
Lamont, Michèle. 2012. “Toward a Comparative Sociology of Valuation and
Evaluation.” Annual Review of Sociology forthcoming.
Lamont, Michèle and Virág Molnár. 2002. “The Study of Boundaries in the
Social Sciences.” Annual Review of Sociology 28:167–195.
Lang, L.-Maurice. 1918-1929. Annuaire des ventes de tableaux, dessins,
aquarelles, pastels, gouaches, miniatures. Paris: Maurice L. (11 vol.).
Level, André. 1959. Souvenirs d’un collectionneur. Paris: A. C. Mazo.
Loeb, Pierre. 1946. Voyages à travers la peinture. Paris: Bordas.
Mainardi, Patricia. 1993. The End of the Salon: Art and the State in the
Early Third Republic. Cambridge: Cambridge University Press.
Marriott, McKim. 1959. “Interactional and Attributional Theories of Caste
Ranking.” Man in India 34:92–107.
Marriott, McKim. 1968. “Caste Ranking and Food Transactions: A Matrix
Analysis.” In Structure and Change in Indian Society, edited by Milton
Singer and Bernard S. Cohn, pp. 133–171, Chicago. Aldine Publishers &
Co.
Moulin, Raymonde. 1987. The French Art Market. New Brunswick: Rutgers
University Press.
Mulkay, Michael and Elizabeth Chaplin. 1982. “Aesthetics and the Artistic Career: A Study of Anomie in Fine-Art Painting.” The Sociological
Quarterly 23:117–138.
Nowicki, Krzysztof and Tom A. B. Snijders. 2001. “Estimation and Prediction of Stochastic Blockstructures.” Journal of the American Statistical
Association 96:1077–1087.
Pachucki, Mark A. and Ronald Breiger. 2010. “Cultural Holes: Beyond
Relationality in Social Networks and Culture.” Annual Review of Sociology
36:205–224.
41
Pachucki, Mark A., Sabrina Pendergrass, and Michèle Lamont. 2007.
“Boundary Processes: Recent Theoretical Developments and New Contributions.” Poetics 35:331–351.
Podolny, Joel M. 1993. “A Status-Based Model of Market Competition.”
American Journal of Sociology 98:829–872.
Podolny, Joel M. 2001. “Networks as the Pipes and Prisms of the Market.”
American Journal of Sociology 107:33–60.
Podolny, Joel M. 2005. Status Signals. A Sociological Study of Market Competition. Princeton: Princeton University Press.
Rengers, Merijn and Olav Velthuis. 2002. “Determinants of Prices for Contemporary Art in Dutch Galleries, 1992-1998.” Journal of Cultural Economics 26:1–28.
Ring, Grete. 1931. “Der junge Künstler und sein Händler in Paris und
Berlin.” Kunst und Künstler 29:179–189.
Rosa, Jose A., Joseph F. Porac, Jelena Runser-Spanjol, and Michael S.
Saxon. 1999. “Sociocognitive Dynamics in a Product Markets.” Journal
of Marketing 63:64–77.
Sagot-Duvauroux, Dominique, Sylvie Pflieger, and Bernard Rouget. 1992.
“Factors Affecting Price on the Contemporary Art Market.” In Cultural
Economics, edited by Ruth Towse and Abdul Khakee, Berlin. Springer.
Schönfeld, Susanne and Andreas Reinstaller. 2007. “The Effects of Gallery
and Artist Reputation on Prices in the Primary Market for Art: A Note.”
Journal of Cultural Economics 31:143–153.
Spence, Michael. 1974. Market Signaling: Informational Transfer in Hiring
and Related Screening Processes. Cambridge: Harvard University Press.
Stark, David. 2009. The Sense of Dissonance: Accounts of Worth in Economic Life. Princeton: Princeton University Press.
Tilly, Charles. 1998. Durable Inequality. Berkeley: University of California
Press.
Turpin, Georges. 1929. La stratégie artistique. Précis documentaire et pratique suivi d’opinions recueillies parmi les personnalités du monde des
arts et de la critique. Paris: Editions de l’Epi.
42
Uzzi, Brian. 1999. “Embeddedness in the Making of Financial Capital: How
Social Relations and Networks Benefit Firms Seeking Finance.” American
Sociological Review 69:319–344.
Uzzi, Brian and Ryon Lancaster. 2004. “Embeddedness and Price Formation
in Corporate Law Markets.” American Sociological Review 69:319–344.
Velthuis, Olav. 2003. “Symbolic Meanings of Prices: Constructing the Value
of Contemporary Art in Amsterdam and New York Galleries.” Theory &
Society 32:181–215.
Velthuis, Olav. 2005. Talking Prices. Symbolic Meaning of Prices on the
Market for Contemporary Art. Princeton: Princeton University Press.
Vollard, Ambroise. 1936. Recollections of a Picture Dealer. Translated from
the Original French Manuscript by Violet M. MacDonald and Now First
Published in any Language. Boston: Little Brown.
Weill, Berthe. 1933. Pan ! . . . Dans l’œil ! ou trente ans dans les coulisses
de la peinture contemporaine. Paris: Librairie Lipschutz.
Wheat, Christopher. 2008. “Multimodel Identification of Group Structure
in Network Data.” MIT Sloan School of Management Working Paper.
White, Harrison C. and Cynthia White. 1965. Canvases and Careers. Institutional Change in the French Painting World. New York: John Wiley &
Sons Inc.
Zelizer, Viviana. 1979. Morals and Markets: The Development of Life Insurance in the United States. New York: Columbia University Press.
Zelizer, Viviana. 1981. “The Price and Value of Children: The Case of
Children’s Insurance.” American Journal of Sociology 86:1036–1056.
Zuckerman, Ezra W. 1999. “The Categorical Imperative: Securities Analysts
and the Illegitimacy Discount.” American Journal of Sociology 104:1398–
1438.
Zuckerman, Ezra W. 2004. “Structural Incoherence and Stock Market Activity.” American Sociological Review 69:405–432.
43
Appendix A. Blockmodeling procedure
If the relational purity and consecration power of galleries are to influence
artists’ prices, the relational pattern of cross-representations across galleries
should display a clear structure to observers of the market for modern art.
Galleries, in other words, should clearly arrange according to a principle
of greater or lesser relational purity. Blockmodeling procedures seem wellsuited to capture this overall arrangement. Yet classic blockmodeling algorithms – such as concor for example (Breiger et al., 1975) – do not provide
rigorous measures of the clarity of network patterns I am after here (see
Doreian et al., 2005; Hsieh and Magee, 2008; Wheat, 2008).
The stochastic approach to blockmodeling introduced by Nowicki and
Snijders (2001) overcomes this limitation. Given a set of underlying relational data and a number of structural equivalence positions, it uses a
simulation approach to determine the probability that two individuals belong to the same position, and the probability distribution of relations across
positions. Most critically, it also offers various statistics for identifying the
blockmodel – if any – that best captures the structure of the underlying
data. This permits ascertaining both whether the observed network displays
a clear-cut structure and, if so, what that structure is. Two chief statistics
are of interest here. The information statistic shows how much of the information in the initial data is restituted by a given blockmodel. This statistic
is 0 if the relations between pairs of vertices are entirely predicted by the respective structural equivalence positions they belong to. The higher it is, in
contrast, the less these positions tell us about the actual relations between
vertices. A better blockmodel therefore has a lower information statistic.
The clarity statistic, on the other hand, measures for a given number of
positions our certainty that any two individuals belong to the same position
– or to different positions. In other words, it indicates the propensity of a
n-position partition to unequivocally summarize the positions of vertices in
the underlying network. The clarity statistic is 0 if, for any pair of vertices,
we know for sure that they are approximately structurally equivalent – or
not. It is 1 if any pair of vertices has a .5 probability of being formed of
individuals belonging to the same position. Likewise, then, the smaller this
statistic, the better the blockmodel.
Table 7 reports values of the information and clarity statistics for various blockmodels and for two specifications of ties in the network of crossrepresentations of artists across the 120 galleries. In the upper panel, a tie
between galleries is defined by their sharing of at least one artist. Five galleries, which did not share any artist with other galleries, are here excluded
44
from the analysis. In the lower panel a tie is defined as the sharing of two
artists or above, which excludes eighteen galleries. Lines in bold signal the
blockmodel that the combination of both statistics indicates as most relevant. Information and clarity are here interpreted both in terms of levels
and trends – that is, taking into account the marginal gain obtained by
adding one position to the blockmodel. The very low values of the clarity
statistic show that the market exhibited a remarkably clear relational structure. Examining the most relevant partitions for various specifications of
ties also reveals many regularities: the number and composition of positions
are relatively similar regardless of the dichotomization threshold, suggesting
robust results.
Table 7: Statistics used in determining the optimal blockmodel,
for two specifications of ties between galleries
Tie definition
Tie = 1 or more
shared artists
Tie = 2 or more
shared artists
Number of
positions
Information
Clarity
2 positions
3 positions
4 positions
5 positions
6 positions
7 positions
8 positions
2 positions
3 positions
4 positions
5 positions
6 positions
7 positions
8 positions
.465
.420
.390
.367
.347
.334
.320
.431
.393
.369
.348
.321
.312
.306
.316
.188
.150
.178
.078
.154
.128
.090
.145
.125
.078
.058
.060
.073
The remainder of this article focuses on the six-position blockmodel obtained when a tie between galleries was defined as two or more shared artists.
Several reasons account for this choice. Theoretically, first, this specification
of ties between dealers seems reasonable to capture meaningful affiliations
45
between galleries. The sharing of one painter could more easily be interpreted by observers as the result of chance. This definition of ties also yields
the least equivocal blockmodel according to the clarity statistic. Finally, on
the more empirical side, a closer examination of the number and composition of positions in the best blockmodels for the two specifications of ties
shows that they yield largely consistent results.
46
Appendix B. Auction prices as an indicator of the
economic value of artists
For the bulk of artists auction prices fall well below prices asked for their
work on the primary market.18 Figure 4 for example plots the highest and
lowest prices asked by gallery Percier for works by three of its young artists
– Francisco Borès, Irène Lagut, and Léon Zack – on the occasion of their
solo exhibitions, against the highest and lowest prices commanded by their
works at auctions over the same period. Size of the works was not reported
in the exhibition catalogs, so that the comparison is somewhat rough. Yet
it shows that for all three artists, the range of auction prices falls short of
reaching even the lower range of gallery prices – despite the fact that the
works sold at auctions, especially by Lagut and Borès, were indeed relatively
large.
3500
3000
2500
2000
1500
1000
500
0
Francisco Borès
1926‐27
Irène Lagut
1927‐28
Léon Zack
1927‐28
Figure 4: Gallery and auction prices over one season for three artists
of the Percier gallery. Full dots indicate the range of prices as posted in the
gallery’s exhibition catalogs, empty dots the range of selling prices at auctions.
Source: Fonds Galerie Percier, Bibliothèque Kandinsky, Paris, and Lang (1918-1929).
Various mechanisms can explain why gallery prices range systematically
higher than auction prices for comparable works by the same artist. Most
obviously, as DiMaggio and Louch (1998) have shown in other instances,
transactions mediated by personal relations, as is the case when a dealer
18
On this issue, see in particular Hutter et al. (2007).
47
is involved, are likely to entail an element of trust that auction sales do
not provide, and for which customers may be willing to pay a monetary
premium.
Gallery prices can be in better accordance with auction prices when
artists are more established. This seems to be the case of Henri Matisse’s
paintings throughout the late 1910s and early 1920s (Figure 5). When an
artist is in high demand, the difference between his auction and gallery prices
can even be reversed, with the latter falling behind. Rather than sticking
to the auction rate, a dealer can indeed prefer to sell for a lesser price to
carefully targeted amateurs, such as those he deems strategic to an artist’s
career. In a 1917 letter to Henri Matisse, Felix Fénéon, the manager of the
Bernheim-Jeune galleries, for example writes:
Our Lausanne branch has sold two of your paintings, for 11,500
French francs, to Mr. [Josef] Mueller, a Solothurn manufacturer
of screws for watches and eyeglasses. [. . . ] This is a very cheap
deal. Yet for the first time, to my knowledge, your work enters a Swiss collection, and it was important to make this first
step. [. . . ] Switzerland has a handful of amateurs of good painting. We hope they will follow Mr. Mueller’s example: if so, we
shall no more resort to the concessions we had to make for this
debut (Archives Matisse, Issy-les-Moulineaux, Bernheim-Jeune
Correspondence, item 170821a).
Overall however, there is certainly a correlation between prices on the
primary market and at auctions – hence the findings by Beckert and Rössel
(2004) showing relatively similar patterns in the determinants of the ones
and the others. This correlation can be observed both between artists and
over time for the same artist, as evidenced by Figure 5 in the case of Matisse. Practically, it arises from the shared understanding that public prices
breed the expectations of collectors regarding over-the-counter prices on the
primary market. As Turpin (1929) puts it,
The first valuation of an artist takes place when a painting appears at Hôtel Drouot, and generally rests on the gallery price.
The dealer will sustain this price using all possible means. From
this day on artist and dealer are tied to one another. If he wishes
to, the dealer can even have a few paintings by the artist he wants
to establish a value for auctioned off. All he needs to do is add
them to the listing of an existing sale. Obviously this strategy
48
5
4
3
2
1
0
Figure 5: Selling prices per square centimeter for paintings by Henri Matisse.
Full dots indicate prices in galleries, empty dots prices at auctions. Prices have
been adjusted for inflation and are expressed in constant 1917 francs.
Source: Archives Matisse, Issy-les-Moulineaux, and Lang (1918-1929).
involves numerous sacrifices for the dealer, since he must be prepared to absorb all the artist’s production that amateurs willing
to cash in on their collection could throw onto the market. An
artist also has an interest in supporting his prices by himself, in
case his dealer fails to do so, since the latter could as well walk
away if prices happened to plummet brutally. [. . . ] The artistic
strategy thus suggests having one’s works valued through auction sales as early as possible, provided that the initial price is
not too high, and that one does not already have too many works
in private collections.
Upsides of such valuation: amateurs, realizing that they own
more than a mere painting, but instead a real asset – that they
can realize either at auctions or directly with a gallery – will be
less reluctant to buy further pieces by the artist. Gradually, then,
artist and dealer should be in a position to increase the market
value of the artist’s work, which will be recognized by critics and
amateurs, and supported by auctioneers (Turpin, 1929, 102-103,
emphasis in the original).
The second reason for choosing auction prices as a barometer of the value
of artists – and therefore as a means to explore its determinants – has to do
49
with the evolution of art market institutions themselves in the wake of World
War I. Up until 1914, only a handful of sales organized at Drouot would
revolve around the work of living artists.19 The situation evolved around
1920. Although infamous for having unfairly hit the personal businesses
of Wilhelm Uhde and Daniel-Henry Kahnweiler, the sales of these German
dealers’ collections and stocks, which had been seized at the outbreak of
war, also paradoxically spread the notion that modern art could encounter
demand at the auction house. Indeed, however flooded the market was with
their work as a consequence of the sales, which took place between May 1921
and May 1923, some artists (in particular Derain, Vlaminck, Van Dongen,
and to a lesser extent Picasso) pulled off relatively high and steady prices.20
The growing legitimacy of modern art as a marketable product prompted
Drouot auctioneers to organize more sales featuring modern works alone.
Figure 6 thus shows the evolution of the number of modern art specific
sales at Drouot over the late 1910s and 1920s. Mixed sales only are taken
into account (as opposed to those featuring the estate of single collectors).
There were ten such sales in season 1918-19, and 27 in season 1927-28. As
can be seen, despite a drop around 1922, probably attributable to the still
depressing effect of the Kahnweiler sales, the growth of modern art specific
sales largely exceeded that of other types of sales. As a market category,
in other words, modern art underwent a breakthrough in the 1920s. The
new category furthermore had its recognized officiant, auctioneer Alphonse
Bellier, who at the end of the 1920s presided over about half of auction sales
specifically devoted to modern art.21
Whether the rise of modern art as an auction category contributed to
the general increase in its prices throughout the 1920s is a question beyond
the scope of this article.22 More important here is the fact that by the end
of the decade, Drouot had become a global exchange for modern painting,
where works by living artists would be traded on a regular basis. “Hôtel
19
The sale of La Peau de l’Ours, held on March 2, 1914, was no doubt the most famous
of such sales, probably because it was unexpectedly successful, and helped propagate the
idea that cutting-edge painting could be a profitable investment. On the Peau de l’Ours
society, and on the sale itself, see in particular the account provided by the society’s
manager in Level (1959).
20
For a detailed analysis of the sales, see Gee (1981, Appendix F, 19-32). On the positive
impact of the sales for modern art, also see Level (1959, 71-72).
21
On Bellier, see Turpin (1929, 108-110), Moulin (1987, 18).
22
For an investigation of this very question in the case of contemporary Indian art,
see Khaire and Wadhwani (2009), who build on the recent scholarship addressing sociocognitive categories and valuation in markets (e.g. Espeland and Stevens, 1998; Rosa et al.,
1999; Zuckerman, 1999, 2004).
50
300
250
200
Modern paintings
sales
All auction sales
150
100
50
Figure 6: Auction sales and auction sales specifically devoted
to modern painting at Hôtel Drouot, 1918-19 to 1928-29.
Season 1918-19 = 100.
Source: Lang (1918-1929).
Drouot is the stock exchange of modern art”, Turpin thus observes in 1929
(Turpin, 1929, 107; also see Basler, 1926). And as early as 1925, Picasso’s
dealer Paul Rosenberg writes to his artist:
I have never had so much to do, every collector in the world
is in Paris [. . . ] The Gangnat sale [held at Drouot on June 24
and 25, 1925] was a triumph, and people are getting crazy. [. . . ]
Painting has become an exchange, it is incredible how an universal atmosphere has emerged around French painting (Paul
Rosenberg – Pablo Picasso correspondence, July 7, 1925, Musée
National Picasso, Paris).
The auction market for modern painting, as a consequence, became large,
deep and reactive enough in the late 1920s – and it attracted bidders from
enough horizons – that it can be deemed a good place to look for an indicator
of the value of artists.
51
Appendix C. Outcome of the survey by art journal
L’Art vivant, 1925
Artist
Matisse Henri
Derain André
Dunoyer de Segonzac André
Bonnard Pierre
Maillol Aristide
Picasso Pablo
Utrillo Maurice
Braque Georges
Vlaminck Maurice
Rouault Georges
Vuillard Edouard
Dufresne Charles*
Denis Maurice
Friesz Othon
Marquet Albert
Dufy Raoul
Moreau Luc-Albert
Favory André
Laurencin Marie
Léger Fernand
Van Dongen Kees
Signac Paul
De Waroquier Henri
Guérin Charles
Laprade Pierre
Lhote André
Besnard Albert
Boussingault Jean-Louis
Flandrin Jules
Forain Jean-Louis
Le Fauconnier Henri
Léopold-Lévy
Lurçat Jean
Marval Jacqueline
Modigliani Amedeo
Puy Jean
Roussel Ker-Xavier
Simon Lucien
Valadon Suzanne
Vallotton Félix
Alix Yves
Charlot Louis
Charmy Emilie
Cottet Charles
Daragnès Jean-Gabriel
De la Fresnaye Roger
Desvallières Georges
Dufrénoy Georges
Girieud Pierre
Gromaire Marcel
Hervieu Louise
Laboureur Jean-Emile
Laurens Henri
Sex
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
F
M
M
M
M
M
M
M
M
M
M
M
M
M
M
F
M
M
M
M
F
M
M
M
F
M
M
M
M
M
M
M
F
M
M
Born
1869
1880
1884
1867
1861
1881
1883
1882
1876
1871
1868
1876
1870
1879
1875
1877
1882
1888
1883
1881
1877
1863
1881
1875
1875
1885
1849
1883
1871
1852
1881
1882
1892
1866
1884
1876
1867
1861
1865
1865
1890
1878
1877
1863
1886
1885
1861
1870
1876
1892
1878
1877
1885
Dead
1954
1954
1974
1947
1944
1973
1955
1963
1958
1958
1940
1938
1943
1949
1947
1953
1948
1937
1956
1955
1968
1935
1970
1939
1931
1962
1934
1943
1947
1931
1946
1966
1966
1932
1920
1960
1944
1945
1938
1925
1969
1951
1974
1925
1950
1925
1950
1942
1940
1971
1954
1943
1954
Votes
26
20
19
18
18
18
15
14
13
12
11
9
8
8
8
7
7
6
6
6
6
5
4
4
4
4
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
Artist
Lebasque Henri
Luce Maximilien
Mainssieux Lucien
Marchand Jean
Metzinger Jean
Naudin Bernard
Piot René
Aman-Jean Edmond
Asselin Maurice
Bissière Roger
Blanchard Maria*
Blanche Jacques-Emile
Bouche Georges
Boutet de Monvel Bernard
Caro-Delvaille Henri*
Chavenon Roland
Chéret Jules*
Clairin Pierre-Eugène
Dauchez André
De Dardel Nils*
Déziré Henri
Fautrier Jean
Fix-Masseau Pierre Félix
Foujita Léonard
Fournier Gabriel
Galanis Démétrius
Gleizes Albert
Goerg Edouard
Gris Juan
Guénot Auguste
Herbin Auguste
Heuzé Edmond
Huyot Albert
Kisling Moïse
Kvapil Charles
Lipchitz Jacques
Lotiron Robert
Manguin Henri
Mare André
Martin Henri
Ménard René
Miró Joan
Muter Mela
Ottmann Henri
Pascin Jules
Péquin Charles
Quizet Alphonse
Sabbagh Georges-Hanna
Schuffenecker Emile*
Simon-Lévy*
Valmier Georges*
Valtat Louis
Willette Adolphe*
Sex
M
M
M
M
M
M
M
M
M
M
F
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
F
M
M
M
M
M
M
M
M
M
M
* indicates that the artist was not represented permanently in any of the galleries surveyed in 1928-29.
52
Born
1865
1858
1885
1882
1883
1876
1869
1858
1882
1886
1881
1861
1874
1884
1876
1895
1836
1897
1870
1888
1878
1898
1869
1886
1893
1882
1881
1893
1887
1882
1882
1884
1872
1891
1884
1891
1886
1874
1885
1860
1862
1893
1876
1877
1885
1879
1885
1887
1851
1886
1885
1869
1857
Dead
1937
1941
1958
1941
1956
1946
1934
1936
1947
1964
1932
1942
1941
1949
1928
NA
1932
1980
1948
1953
1965
1964
1937
1968
1963
1966
1953
1969
1927
1966
1960
1967
1968
1953
1957
1973
1966
1949
1932
1943
1930
1983
1967
1927
1930
1963
1955
1951
1934
1973
1937
1952
1926
Votes
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1