American Economic Association
Art and Money
Author(s): William N. Goetzmann, Luc Renneboog and Christophe Spaenjers
Source: The American Economic Review, Vol. 101, No. 3, PAPERS AND PROCEEDINGS OF THE
One Hundred Twenty Third Annual Meeting OF THE AMERICAN ECONOMIC ASSOCIATION
(MAY 2011), pp. 222-226
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/29783743 .
Accessed: 26/02/2014 07:27
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.
http://www.jstor.org
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions
American Economic Review: Papers & Proceedings 2011, 101:3, 222-226
http://www.aeaweb. org/articles.php ?doi= 10.1257/aer. 101.3.222
Art and Money
By William N. Goetzmann, Luc Renneboog, and Christophe Spaenjers*
Beauty is commonly a gratification of our
sense of costliness masquerading under
thename of beauty.
?Thorstein Vehlen
Unless cast inplatinum and covered with dia?
monds, as in the case of a 2007 Damien Hirst
sculpture, a work of arthas little intrinsicvalue.
Nevertheless, works of arthave fromtime to time
fetched shockingly high prices, at least from
the perspective of ordinary wage earners. The
highest amounts have been paid for creations
of deceased artists,but also living artists?Hirst
being the exemplar?have commanded multi
million dollar or pound sums for theirwork. It
is still largely a puzzle what determines these
prices and theirpattern over time.
Yet it is clear that theprice of an artobject is
limited only by the amount that collectors are
willing and able topay for it.Given the interest
of many high net worth individuals in art,we
analyze the impacton artprices of timevariation
inhow much money thewealthiest members of
society can spend.
One way tomeasure
changes in wealthy indi?
viduals' buying power is to look at stockmarket
returns.Equities are typically held more widely
among themost affluent.A number of studies
have indeed looked at therelation between stock
market and artmarket trends.1In thispaper, we
extend thiswork over amuch longer time frame,
startingour study in the firsthalf of the nine?
teenth century.
A complementary approach to proxying for
collectors' ability to purchase art consists of
studying theevolution of top incomes over time,
especially if thehighest incomes also go to the
wealthiest individuals.We thereforeempirically
investigate the linkbetween the income distribu?
tionon theone hand and artprices on theother,
a relationship which has not been analyzed
before.2
I. Data
In this section, we firstconstruct a long-run
art price index. Since the index is largely based
on London sales and is expressed in British
pounds (GBP), we also collect equity market
and income data forGreat Britain. Insofar as it
was mostly British individuals who bought the
considered artists at British auctions over our
time frame, thisprocedure seems justified.
A. Art Prices
We start by building a long-term art price
index.To do so,we go back to the auction sales
data collected by Gerard Reitlinger (1961),
who investigated the history of the British
paintings and drawings market. Despite the*Goetzmann: Yale School ofManagement, 135 Prospect
Street,New Haven, CT 06511 (e-mail: william.goetzmann@
yale.edu); Renneboog: CentER, Tilburg University, P.O.
Box 90153, Tilburg, 5000 LE, theNetherlands (e-mail: lue.
renneboog@uvt.nl); Spaenjers: CentER, Tilburg University,
P.O. Box 90153, Tilburg, 5000 LE, theNetherlands (e-mail:
c.spaenjers@uvt.nl). The authors would like to thank
Anthony Atkinson, Fabio Braggion, Elroy Dimson, Marc
Goergen, Richard Grossman, Benjamin Mandel, Kim
Oosterlinck, Emmanuel Saez, Darius Spieth, Radomir
Todorov, John Turner, and seminar participants at Cardiff
Business School, Erasmus University Rotterdam, Queen's
University of Belfast, Tilburg University, and the Financial
History Workshop at the University of Antwerp for help?
ful comments and data. Spaenjers thanks theNetherlands
Organisation for Scientific Research (NWO) for financial
support. Part of this paper was written while Spaenjers
was visiting Columbia University, whose hospitality is
appreciated.
1
For example, William N. Goetzmann (1993) docu?
ments a lagged relationship between art prices and the
stock market. Olivier Chanel (1995) presents evidence that
stock markets Granger-cause art prices. Takato Hiraki et al.
(2009) show thatpositive wealth shocks to Japanese inves?
tors affected theirart purchases in the 1980s, lifting theprice
level in the global artmarket. While the latter authors treat
art as a luxury consumption good, Benjamin R. Mandel
(2009) constructs a model inwhich a positive correlation
between equity returns and art returns is induced by the use
of art as a savings asset.
2
In contrast, in the real estate literature, Joseph Gyourko,
Christopher Mayer, and Todd Sinai (2006) and Stijn Van
Nieuwerburgh and Pierre-Olivier Weill (2010) have recently
acknowledged the importance of the distribution of income
in determining housing price levels.
222
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions
VOL. 101 NO. 3 ART AND MONEY 223
well-documented selection issues with the
Reitlinger data (Guido Guerzoni 1995), they
still constitute a unique historical overview
of auction sales since the eighteenth century.
The artistswhose sales are listed in this source
mostly conform to English standards of taste.
All transactionprices are expressed inGBR
Reitlinger's data have previously been used
to estimate the returns on art by, among oth?
ers,William J.Baumol (1986) and Goetzmann
(1993). In linewith these studies, we identify
all repeated sales within Reitlinger's book. This
gives us a dataset of 1,096 sales pairs until 1961,
excluding buy-ins (i.e., items for which the
reserve price has not been met).
We then look up all 6,661 works listed in
Reitlinger (1961) in the dataset constructed
by Luc Renneboog and Christophe Spaenjers
(2009), which contains more than one million
auction sales until 2007, and try to identify
resales of those sameworks inGreat Britain.We
treata transaction as a resale only when there is
a unique match of a nonambiguous title,which
occurs in253 cases.3 In totalwe thusend upwith
a dataset containing 1,349 repeated sales. Since
thedata are very sparse for thefirstdecades cov?
ered by Reitlinger, we delete the 13 pairs for
which thepurchase occurred prior to 1765. This
leaves us with 1,336 repeated sales.
To estimate a price index, we follow the
Bayesian formulation of a repeat sales regres?
sion,which imposes some additional restrictions
on theestimation, outlined inGoetzmann (1992,
1993). The Bayes formulation avoids spurious
negative autocorrelation in the estimated return
series and is particularly useful when thenum?
ber of observations is relatively small. Prior to
applying theregression toour dataset,we deflate
all transaction prices to real GBP. More details
on our estimationmethodology can be found in
Goetzmann, Renneboog, and Spaenjers (2010).
We show the time series of the index val?
ues since 1830 in Figure 1. The figure sug?
gests a relationship between the real economy
and art prices. For example, we see significant
1,000.0
omomomom
coldooocoloooo
0000000)0)0)0)0
T-T-T-1-T-T-1-<m
Figure 1.Art Prices inReal GBP 1830-2007
Notes: This figure shows the annual art price index in real
GBP for the period 1830-2007, on a logarithmic scale. The
index value in 1830 is put equal to 1.
price drops duringWorld War I, over theGreat
Depression in the 1930s, and after theoil crisis
in 1973. In contrast,we find strongprice appre?
ciations throughout the 1960s, during the art
market boom at theend of the 1980s, and in the
2000s (at least until 2007). However, art prices
stayed remarkably low formany years in the
middle of the twentieth century. In real terms,
the price level of 1913 was not reached again
until 1968, despite decades of economic growth.
We will henceforward refer to our log price
index as Art. The firstdifferences of this index
constitute our estimates of the log returnson art.
A concernwith any art index is thatsurvivorship
issues can put an upward bias on the estimated
returns (Goetzmann 1996). However, since our
focus isnot on estimating returnsbut on identify?
ingwhat determines thevariation in art returns,
thisdoes not have tobe amajor problem here.
B. Equity Returns and Income Data
We build a historyofBritish stockprice returns
for theperiod 1830-2007, based on Richard S.
Grossman (2002), Graeme G. Acheson et al.
(2009), and Elroy Dimson et al. (2009). We cre?
ate yearly indices covering total return,capital
appreciation, and dividend yield, transformed
into real terms.We call the natural log series
Equities, Equities (cap.), and Equities (div).
A recent literaturehas investigated the evo?
lution of top incomes over the course of the
3
We classify a transaction inRenneboog and Spaenjers
(2009)
as a match to a sale in Reitlinger's list ifwe find
strong evidence of the existence of only one work with the
same titleby the same artist.Also, we exclude objects with
attribution classifications and with very general titles (or
titles that point to a much-used subject of the artist), and
objects thatwent tomuseums according toReitlinger.
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions
224 AEAPAPERSANDPROCEEDINGS MAY2011
twentieth century.We use data from A. B.
Atkinson and T. Piketty (2010), who themselves
rely on tax data, to build a consistent series of
the share of total income received by the top0.1
percent of all income earners in theUK for the
period 1908-2005.4 This series will be referred
toas Inequality. Interestingly,income inequality
generally decreased through the firsthalf of the
twentiethcentury and has increased over the last
few decades, which is roughly in linewith the
patternobserved forartprices.We also construct
the series Income and Top income,which mea?
sure the logs of deflated total personal income
and deflated income of the top0.1 percent earn?
ers in every year.
II. Results
We firstpresent the results of comovement
regressions that relate art returns to equity
returnsand changes in the income distribution.
Thereafter,we undertake a cointegration analy?
sis to investigatewhether we can identifya long
rundriver of artprices.
A. Comovement Regressions
InTable 1,we present the results of ordinary
least squares (OLS) regressions that relate the
log returns on art to equity returns, changes
in total personal income, and changes in the
income distribution. Below each coefficient,
we present theNewey-West standard error that
accounts for heteroskedasticity and first-order
autocorrelation in theerror terms.We also show
thenumber of observations and theR2 foreach
regression.
Model (1) regresses the returns on art on
equity capital growth rates and dividend yields
over the 1830-2007 period. Because of poten?
tial nonsynchroneity between our art price
index (which aggregates information per cal?
endar year) and equity prices, we also include
lagged equity capital returns.Model (2) relates
art returns to thegrowth of personal income and
changes in income inequality. Data for these
Table 1?Comovement Regressions
1830 1908 1908
-2007 -2005 -2005
_(1)
(2) (3)
A Equities (cap.) 0.13*0.11
(0.07) (0.07)
A_! Equities (cap.)
0.21*** 0.21***
(0.05) (0.06)
A Equities (div.) -0.01
(0.21)
A Income 0.14 0.20
(0.26) (0.26)
A Inequality 14.35*** 9.59***
(4.18) (4.13)
Observations 176 97 96
R2 0.13 0.12 0.23
Notes: The returns on art are regressed on a constant and
a changing set of independent variables, listed in the first
column. All models are estimated using OLS. Newey-West
standard errors are inparentheses below the coefficients.
***
Significant at the 1percent level.
**
Significant at the 5 percent level.
*
Significant at the 10 percent level.
variables are only available over the 1908-2005
time frame. In model (3), we check how the
results on the income variables change once we
control for equity returns. In this last specifica?
tion,we exclude thedividend variable, because
dividends should already be captured by theper?
sonal income variables.
The results inTable 1 show thatequity capital
growth, and especially lagged stock price appre?
ciation, has a statistically and economically sig?
nificant impact on artprices. However, we also
find thatartprices risewhen inequality goes up,
even when controlling forequitymarket trends.
The coefficient on Inequality inmodel (3) sug?
gests that a 1 percentage point increase in the
share of totalpersonal income earned by the top
0.1 percent triggersan increase in art prices of
about 10 percent. The R2 of thismodel is sub?
stantially higher than when considering the
impact of equities or the income distribution
separately.
B. Cointegration Analysis
The above evidence on comovement between
equity markets and income inequality on the
one hand and artmarkets on the other is based
on relatively short-termeffects. The long-term
4
Data on the top 0.1 percent income share are missing
for a limited number of years. For the periods 1908-1912
and 1987-1992, we impute the share of the top 0.1 percent,
based on the available income shares. For the years 1961
and 1980, we linearly interpolate the top 0.1 percent income
share.
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions
VOL. 101 NO. 3 ART AND MONEY 225
Table 2?Cointegration Tests
Trace Max. eigenvalue
Equities
Equities (cap.)
GDP
Income
Top income 20.27***
5.67
8.28
7.29
10.18
5.66
8.27
7.25
9.54
17.82**
Notes: This panel shows the results of Johansen's cointegra?
tion tests over the period 1908-2005. In each case, the null
hypothesis is that of no cointegrating relation with Art. No
trend is assumed in the cointegrating equation. The test sta?
tistics of both the trace and themaximum eigenvalue tests
are reported.
***Significant at the 1percent level.
**
Significant at the 5 percent level.
*
Significant at the 10 percent level.
nature of our data series (and the fact that the
series are integratedof order one) allows further
exploration of the factors that drive art prices
over the long run. If it is really thehigh-income
individuals who determine theprice level in the
artmarket, then one would expect Top income
(but not necessarily Income) to be cointegrated
with artprices.
Table 2 shows the results of Johansen's coin?
tegration tests applied to our time series since
1908.We report the results of trace and maxi?
mum eigenvalue tests,assuming no trend in the
cointegrating equation and including one lagged
firstdifference in themodel.
We find that the null hypothesis of no coin?
tegrationbetween Art and the series in the first
column can never be consistently rejected,
except in thecase of Top income.5Over the long
run, the income of thewealthy, or at least of the
highest earners, seems a key factor in theprice
formation in theartmarket.
Profound changes have taken place in the art
market since themiddle of theprevious century.
Without doubt, the artmarket has become more
globalized, or at least reached the level of inter?
national integration it enjoyed in the late nine?
teenthcentury.One may thusexpect therelation
between ourUK artprice index on theone hand
III. Analysis per Subperiod
and theBritish equity market and income dis?
tributionon the other to be weaker afterWorld
War II than before. Therefore, we repeat our
main analyses, but differentiate between the
1908-1945 and thepost-1945 period.
The results of the comovement regressions,
which are reported inGoetzmann, Renneboog,
and Spaenjers (2010), indicate that lagged
British equity capital growth has a similar posi?
tive impact on our artprice index inboth subpe
riods. In contrast, the previous findings on the
impact of income inequality on artprices seem
attributable to trends in theearly twentiethcen?
tury,when substantial decreases in inequality
eroded the relative buying power of thewealthi?
est. Our results are consistent with thehypoth?
esis that the income concentration in Great
Britain mattered less in the second half of the
twentiethcentury indetermining artprices.
Still, even for the postwar period, we can
reject the hypothesis that top incomes are not
cointegratedwith artprices.
IV. Conclusion
This articlehas investigatedhow equity returns
and personal income?more generally, money?
determines theprice of art.We are able toconfirm
and strengthenprevious evidence thatequitymar?
ketmovements affect art prices, using a newly
constructed art price index. We find weaker
evidence for the impact of income inequal?
ity.Although there is evidence that changes in
income inequality had an important effect on
British artprices in thefirsthalf of the twentieth
century,and that thiseffect is significantfor the
overall time frame,we do not confirm the result
for the postwar period.We conjecture that this
may be due to theglobalization in thedemand for
high-quality art in the later era. Arguably more
important,however, is thatwe find cointegrat
ing relationships between top incomes and art
prices, both for the complete 1908-2005 period
and since 1945. These relationships support the
Veblenian view of art as an instrumentof social
competition among theveryrich.
5
We find similar results when allowing for a linear trend
in the cointegration equation or when including lagged
equity capital growth as an exogenous variable.
REFERENCES
Acheson, Graeme G., Charles R. Hickson, John
D. Turner,and Qing Ye. 2009. "Rule Britannia!
British Stock Market Returns, 1825-1870."
Journal ofEconomic History, 69(4): 1107-37.
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions
226 AEAPAPERSANDPROCEEDINGS MAY2011
Atkinson, A. B., and T. Piketty, ed. 2010. Top
Incomes: A Global Perspective. New York:
Oxford University Press.
Baum?l, William J. 1986. "Unnatural Value: Or
Art Investmentas Floating Crap Game." Amer?
icanEconomic Review, 76(2): 10-14.
Chanel, Olivier. 1995. "Is ArtMarket Behaviour
Predictable?" European Economic Review,
39(3-4): 519-27.
Dimson, Elroy, Paul Marsh, Mike Staunton, and
JonathanWilmot. 2009. Credit Suisse Global
Investment Returns Yearbook 2009. Zurich:
Credit Suisse Research Institute.
Goetzmann, William Nelson. 1992. "The Accu?
racy ofReal Estate Indices: Repeat Sale Esti?
mators." Journal of Real Estate Finance and
Economics, 5(1): 5-53.
Goetzmann, William N. 1993. "Accounting for
Taste: Art and the Financial Markets over
Three Centuries." American Economic Review,
83(5): 1370-76.
Goetzmann,William N. 1996. "How Costly Is the
Fall fromFashion? Survivorship Bias in the
Painting Market." In Economics of theArts:
Selected Essays. Vol. 237, Contributions to
Economic Analysis, ed. Victor A. Ginsburgh
and Pierre-Michel Menger, 71-84. New York:
Elsevier.
Goetzmann, William N., Luc Renneboog, and
Christophe Spaenjers. 2010. "ArtandMoney"
Yale InternationalCenter forFinanceWorking
Paper 09-26.
Grossman, Richard S. 2002. "New Indices of
British Equity Prices, 1870-1913." Journal of
Economic History, 62(1): 121-46.
Guerzoni, Guido. 1995. "Reflections on His?
torical Series ofArt Prices: Reitlinger's Data
Revisited." Journal of Cultural Economics,
19(3): 251-60.
Gyourko, Joseph, Christopher Mayer, and Todd
Sinai. 2006. "Superstar Cities." National
Bureau ofEconomic Research Working Paper
12355.
Hiraki, Takato, Akitoshi Ito, Darius A. Spieth,
and Naoya Takezawa. 2009. "How Did Japa?
nese Investments Influence InternationalArt
Prices?" Journal ofFinancial and Quantitative
Analysis, 44(6): 1489-1514.
Mandel, Benjamin R. 2009. "Art as an Invest?
ment and Conspicuous Consumption Good."
American Economic Review, 99(4): 1653-63.
Reitlinger,Gerald. 1961. The Economics ofTaste:
The Rise and Fall of Picture Prices 1760
1960. London: Barrie and Rockliff.
Renneboog, Luc, and Christophe Spaenjers. 2009.
"Buying Beauty: On Prices and Returns in the
Art Market." Tilburg University CentER Dis?
cussion Paper 2009-15.
Van Nieuwerburgh, Stijn, and Pierre-Olivier
Weill. 2010. "Why Has House Price Disper?
sionGone Up?" Review ofEconomic Studies,
77(4): 1567-1606.
This content downloaded from 147.251.185.122 on Wed, 26 Feb 2014 07:27:01 AM
All use subject to JSTOR Terms and Conditions