Pricing art and the art of pricing: On returns and risk in art auction markets

• Published date: 01 November 2022
• Author: Yuexin Li,Marshall Xiaoyin Ma,Luc Renneboog
• Date: 01 November 2022
• DOI: http://doi.org/10.1111/eufm.12348

DOI: 10.1111/eufm.12348

ORIGINAL ARTICLE

Pricing art and the art of pricing: On returns

and risk in art auction markets

Yuexin Li

1

|Marshall Xiaoyin Ma

2,3

|Luc Renneboog

4

1

School of Applied Economics, Renmin

University of China, Beijing, China

2

Ant Group, China

3

Erasmus University, the Netherlands

4

Department of Finance, Tilburg

University, Tilburg, the Netherlands

Correspondence

Yuexin Li, School of Applied Economics,

Renmin University of China, Beijing

100872, China.

Email: yuexinli@ruc.edu.cn

Abstract

We study price determinants and investment perfor-

mance of art using a vast sample of transactions

worldwide over the past 60 years. We focus on paint-

ings and drawings which have appreciated at a real

(nominal) annual return of 2.49% (6.24%). Higher art

returns are reached for paintings at the high end of the

price distribution, oil paintings, more recent art

movements and transactions by reputable auction

houses. The risk–return trade‐off of paintings under-

performs that of other passion investments. Paintings'

Sharpe ratios are below those of stocks, bonds and gold

but outperform those of commodities and real estate.

Investments in paintings enter the optimal investment

portfolio.

KEYWORDS

art investment, auction, cultural economics, hedonic pricing

model, repeat sales model

JEL CLASSIFICATION

D44, G20, G11, Z11

Eur Financ Manag. 2022;28:1139–1198. wileyonlinelibrary.com/journal/eufm

|

1139

EUROPEAN

FINANCIAL MANAGEMENT

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits

use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or

adaptations are made.

© 2021 The Authors. European Financial Management published by John Wiley & Sons Ltd.

Yuexin Li gratefully acknowledges supports from the National Natural Science Foundation of China (72204257) and

the Fundamental Research Funds for the Central Universities, and the Research Funds of Renmin University of China

(22XNQT46).

[Correction added on 17 September 2022, after first online publication: Grant information has been updated in this

version.]

1|INTRODUCTION

The growth of art markets is driven by increasing art prices and demand. The underlying

drivers of demand are economic development, wealth accumulation and concentration, income

inequality and the emergence of new art‐buying audiences, such as the financial elite of

developing economies, including China, Russia and India. The modern art market has been

growing rapidly. The market was dominated by European auction houses up to the 1950s but is

now a global market with more than 30,000 auction houses worldwide that have auctioned

artworks created by more than 150,000 artists over the past half‐century. The increase in

studies on art as an investment over the past 30 years gives evidence of increasing interest in

the financial performance of the art market. The estimated value of art and collectables held by

high‐net‐worth individuals is approximately USD 1.74 trillion (Deloitte & ArtTactic, 2019).

Global art sales grew from USD 40 billion in 2009 to USD 67 billion in 2018, dipping to USD

64 billion in 2019. The COVID‐19 pandemic reduced the global art auction value to USD

50 billion in 2020. The auction sector currently accounts for 42% of the value of global art sales,

with dealers and galleries accounting for the remaining 58%. Online sales have grown sig-

nificantly over the past year and received a strong boost during the global pandemic, reaching a

record high of USD 12.4 billion in 2020 (Art Basel & UBS, 2021).

New price records for art are being set constantly. The most striking example is of Leonardo

da Vinci's Salvator Mundi, which was sold by Christie's for USD 450.3 million in November 2017

after 20 min of fierce bidding. This auction broke the previous world auction record for the most

expensive painting sold, Interchange by Willem de Kooning, which sold for USD 300 million but

held the record for only 14 months. In addition to the exorbitant amount of money paid for

Salvator Mundi, the return was substantial, having previously been auctioned for a mere GBP 45

by Sotheby's London in 1958. Although doubts about the painting's authenticity persist, it was

authenticated as a bona fide Leonardo da Vinci work at the beginning of the 21st century. Such

exceptionally high prices and returns raise questions about whether investing in art generally

yields superior returns, what determines the price of paintings, the risk–return trade‐off in

modern art markets and whether art belongs in a diversified investment portfolio.

For this study, we collect almost three million auction transactions of paintings over

1957–2016 and apply several methods to calculate the price index series. In our hedonic pricing

models, the variables include artist reputation, physical characteristics of the painting (e.g.,

size, medium, signed and dated), provenance (pedigree, exhibition history, literature and

certification) and transaction characteristics (auction house [branch], seasonality, etc.). We

verify the size of the returns derived from the price index using three‐stage weighted least‐

squares repeat‐sales regressions, adjacent‐period hedonic regressions and quantile regressions.

Our results show that artist reputation, attribution, authenticity (signed and dated), medium,

size, topic, provenance information and the timing and location of the sale all significantly

affect hammer prices. The coefficients of the time dummies enable us to construct a price

index. Constant‐quality art prices increased annually by a moderate 2.49% (6.24%) in real

(nominal) USD over 1957–2016. Our results are lower than those reported by Goetzmann

(1993), Korteweg et al. (2016), Mei and Moses (2002), Renneboog and Spaenjers (2013) and

Spaenjers et al. (2015) but higher than those of Pesando (1993). An important reason is that

Goetzmann (1993), Mei and Moses (2002) and Spaenjers et al. (2015) use selective samples

comprising only higher‐end artworks and, thus, are less representative of the entire art market.

To comprehend the price fluctuations in modern art auction markets, we examine the in-

vestment performance over different holding periods, including bubble and bust periods, as

1140

|

EUROPEAN

FINANCIAL MANAGEMENT

LI ET AL.

well as for various subsamples based on a stratification of price levels for samples based on the

medium (oil and acrylic paintings, watercolours and gouaches and drawings), art movements,

transactions by auction house reputation, transactions categorised by artist nationality, art

market segments (local vs. international) and paintings created in various phases of an artist's

career cycle. We find that both returns and risk in the first three decades of our sample period

(1957–1986) exceed those in the subsequent three decades (1987–2016). Returns and risk are

also higher for oil and acrylic paintings than for watercolours and drawings, for more recent art

movements (e.g., Minimalism and Contemporary,Pop Art and Abstract Expressionism), auctions

organised by the big four auction houses (Sotheby's, Christie's, Bonhams and Phillips), trans-

actions in the international art auction market and paintings created in the last phase of an

artist's career cycle.

Investments in paintings outperform alternative investments in sculptures but underper-

form investments in other types of collectables or passion goods such as investment‐grade

white diamonds, classic cars, premier cru (first growth) red Bordeaux wines, stamps and fine

violins. The return by unit of risk of art investments is lower than that of equities, corporate

bonds and gold, but exceeds that of global commodities and real estate in the United States

(US). As the correlations between paintings on the one hand and stocks and bonds on the other

are negative, we study whether investments in paintings contribute to optimal portfolio allo-

cation. The academic finance literature has focused on art performance (Korteweg et al., 2016;

Lovo & Spaenjers, 2018; Mei & Moses, 2002; Renneboog & Spaenjers, 2013); its financial and

macro‐economic market drivers, such as equity market evolution and income inequality

(Goetzmann et al., 2011); sentiment and hype (Pénasse et al., 2014); gender bias (Adams

et al., 2021; Bocart et al., 2021; Cameron et al., 2019); the impact of colour

(Ma et al., 2021); behavioural anomalies, such as anchoring (Beggs & Graddy, 2009; Graddy

et al., 2015); art market bubbles (Pénasse & Renneboog, 2021) and artists' death as a supply

shock (Pénasse et al., 2021).

While an art investment is expected to yield a combination of a monetary return and

unobservable emotional value (an aesthetic dividend derived from ownership rights), this

study focuses only on the financial performanceofart.InlinewithAshenfelterandGraddy

(2003), we note that the estimated returns on art vary with samples, the method to calculate

price indices and sample period. Early studies used simple estimation methods without

controlling for variables capturing the quality of the artworks (Baumol, 1986;Frey&

Pommerehne, 1989;Stein,1977).

1

More recent studies use either hedonic regressions or

repeat‐sales regressions to estimate the price movements of art and other illiquid assets

(e.g., real estate). The benefit of hedonic models is that they control for quality changes in

the transacted goods by attributing implicit prices to their utility‐bearing characteristics. In

the commonly used time‐dummy variant of hedonic pricing models, all available transac-

tion data are pooled and prices are regressed on a set of value‐determining attributes,

including time indicators. Under the assumption that all omitted characteristics are or-

thogonal to those included, the coefficients on the time dummies account for constant‐

quality price trends over the sample period (Meese & Wallace, 1997). As no observations are

discarded, hedonic regressions efficiently use available data and, therefore, may provide

1

Stein (1977) considers the auctioned objects each year as a random sample of the underlying stock of art (by deceased

artists) and constructs an index based on the yearly average transaction price. Baumol (1986) and Frey and

Pommerehne (1989) calculate the geometric mean return on works that sold at least twice during the considered

timeframe.

LI ET AL.EUROPEAN

FINANCIAL MANAGEMENT

|

1141