ed

Non-Fungible Token Artworks: More Crypto than Art?

Giulio Anselmi

iew
Università Cattolica del Sacro Cuore
giulio.anselmi@unicatt.it

Giovanni Petrella
Università Cattolica del Sacro Cuore

v
giovanni.petrella@unicatt.it

re
This Draft: July 5, 2022

Abstract
er
Non-Fungible Tokens (NTFs) artworks attract relevant trading volume in the secondary
market. Their prices, primarily denominated in Ethereum cryptocurrency, recently
pe
experienced a substantial increase. In this paper, we first investigate whether the market for
crypto art is informationally efficient and, secondly, if crypto art can be considered as an
alternative to real-world art rather than a haven for crypto-money speculative traders. We
find evidence of market inefficiency with long-lasting autoregressive patterns in price
discovery. As for the nature of crypto art as an asset class, our findings support the idea that
ot

crypto art is unrelated to real-world art. Crypto art prices are mostly determined by gains in
the cryptocurrency of denomination, whereas real-world art is mostly unrelated to the
financial markets. We conclude that crypto art is far from being another medium to express
tn

artistic content and is currently just a storing unit for crypto-money, in absence of relevant
alternatives.
rin

Keywords: crypto art, cryptocurrencies, blockchain, Ethereum, Bitcoin, non-fungible tokens,

NFT, art, real-world art, collectibles, market efficiency, asset pricing, safe haven.
ep

JEL Classification Codes: G14, G15

Pr

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 ed
Non-Fungible Token Artworks: More Crypto than Art?

v iew
re
This Draft: July 5, 2022

Abstract
er
Non-Fungible Tokens (NTFs) artworks attract relevant trading volume in the secondary
market. Their prices, primarily denominated in Ethereum cryptocurrency, recently
pe
experienced a substantial increase. In this paper, we first investigate whether the market
for crypto art is informationally efficient and, secondly, if crypto art can be considered
as an alternative to real-world art rather than a haven for crypto-money speculative
traders. We find evidence of market inefficiency with long-lasting autoregressive
patterns in price discovery. As for the nature of crypto art as an asset class, our findings
ot

support the idea that crypto art is unrelated to real-world art. Crypto art prices are mostly
determined by gains in the cryptocurrency of denomination, whereas real-world art is
mostly unrelated to the financial markets. We conclude that crypto art is far from being
tn

another medium to express artistic content and is currently just a storing unit for crypto-
money, in absence of relevant alternatives.

Keywords: crypto art, cryptocurrencies, blockchain, Ethereum, Bitcoin, non-fungible

rin

tokens, NFT, art, real-world art, collectibles, market efficiency, asset pricing, safe haven.
ep

JEL Classification Codes: G14, G15

Pr

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 1. Introduction

ed
Crypto art is the ultimate spillover of the cryptocurrency universe. Its trading is in
constant evolution with several new investors bidding for the ownership of dematerialized
art every day. As for cryptocurrencies, blockchain technology is underneath the surface

iew
of the asset, hence an entire ecosystem of marketplace, apps, and intermediaries
blossomed in the last two years. Although the asset does not present physical features like
real art, the values are very significant, with marketplaces exchanging millions in volume.
Such exchanges appear to operate with no regulation and supervision in place and the risk

v
for market manipulation is real1.

re
Given the increased relevance of crypto art it is worth to investigate the reliability
of the marketplaces where crypto art is traded and the asset class to which crypto art
belongs. Our goal is therefore twofold: first we want to assess whether the market for
er
crypto art is efficient, second and most importantly we intend to understand whether
crypto art and real art share a common ground in terms of asset valuation or if crypto art
is a different asset class.
pe
A non-fungible token (NFT) is a file with attached a certificate of deed made
tamper-proof by blockchain technology and stored in a digital ledger. As for each mined
bitcoin, an NFT keeps tracks of all the previous owners back to the original creator in
ot

support of its authenticity. NFTs file can be either picture or video and, given the
uniqueness of the file, they can be referred to as crypto art. Yet it is important to specify
that the uniqueness of the file is driven by the afore-mentioned certificate of deed not by
tn

the item itself which, unlike real art, can be copied like any other digital asset. The
strength of blockchain technology is the only bulwark that gives moneywise value to
crypto art. Figure 1 provides a brief snapshot of the NFTs market: once accounting for
rin

less than a few millions a month, it recently experienced an extraordinary growth with
ep

1
See “OpenSea admits insider trading of NFTs it promoted”, BBC, September 17, 2021
(https://www.bbc.com/news/technology-58585342) and “NFT trader OpenSea bans insider trading after
employee rakes in profit”, The Guardian, September 16, 2021
Pr

(https://www.theguardian.com/technology/2021/sep/16/nft-trader-opensea-bans-insider-trading-after-
employee-rakes-in-profit)

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 hundreds of millions traded every month. Crypto art currently accounts for almost the

ed
entire NFTs market.

As for this paper's contribution, our aim is to contribute to the understanding of

crypto artworks pricing providing empirical evidence on its informational efficiency and

iew
the relationship with real-world artworks pricing. We are not aware of empirical studies
investigating such questions.

The relationship between NFTs and the cryptocurrency world is fundamental to

understand the fintech ecosystem. Once cryptocurrencies are widely accepted as a

v
payment method and a measure of value, we must figure out which assets denominated

re
in cryptocurrencies will benefit mostly from this inflow of money in the crypto world. If
NFT express just another way of representing artistic content, then the demand for these
tokens will stem from the already existing demand from art, and the phenomenon for
er
crypto art will normalize as soon as the “crypto bonanza” is over.

Looking at real-world artworks pricing, Candela and Scorcu (1997) find that
pe
contemporary art prices increase with inflation. They also find that returns on artworks
are lower than returns on financial assets and in the long-run art prices are unrelated to
financial asset. Higgs and Worthington (2004) reach similar conclusions but also point
out that art investment provides no diversification benefit given the higher risk than stock
ot

market. Renneboog and Spaenjers (2012) find that only high-income consumer
confidence is relevant to predict art prices.
tn

On the other hand, if NFT represent a wallet for storing crypto-money the demand
for this asset originates from the demand for the main asset, the cryptocurrency (see
Dowling (2021a)). In this case we would expect a behavior like the one we may posit
rin

when we observe a large inflow of new money into a specific currency, due to monetary
easing policy or economic-related shift in the overall risk. In this scenario the demand for
the currency increases, the exchange rate strengthen, and government bond yields drops,
ep

since treasury bond is the main asset where to store the money. We have witnessed this
behavior frequently in recent years. Except that for cryptocurrencies we do not have
government bond or any central bank managing the quantity of crypto-money. If this is
Pr

the case, crypto art would rise in value just for the increase in demand for cryptocurrency
and perish once one of the following scenarios happen: either the inflow of money into

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 cryptocurrencies shrinks, or safer player than artist and creator of artistic content, such as

ed
government and big corporate, start to issue cryptocurrency-denominated equity or debt
instruments. Crypto art demand will endure only if the asset, due to its uniqueness, is
perceived as a safer place than the crypto-government bond where to park your crypto-

iew
money.

The remainder of this paper is organized as follows: Section 2 outlines the world
of crypto art, Section 3 describes the sample, presents the research design, and states
testable hypotheses, Section 4 shows the results of our empirical analysis, and Section 5

v
concludes.

re
2. Crypto Art

In a nutshell, crypto art is a way of making digital art one-of-a-kind. Beside of that, the
definition is still evolving, and the future of the asset is yet to be determined. Crypto art
er
uniqueness originates form the blockchain technology, the same technology that gives
uniqueness to each cryptocurrency coin is now in circulation. Each crypto art has a unique
pe
ID number, and this ID number is checked against a distributed database, that keeps an
immutable record of any transaction. Since its first sale from the creator to the early
collector the change in ownership is written down in the database, and the same happens
for the following change of hands. When an NFT transaction takes place, the data is
ot

timestamped and must be validated across the entire blockchain: every “block” of the
“chain” of transaction is verified and the new block is added on top. The history of all
tn

transactions traces back ownership to the original creator that digitally “signed out” the
art piece guaranteeing for its authenticity. The digital file by itself is no different than any
other .jpg o .bmp file, if allowed can be printed an infinite time without losing any quality
rin

or differences from the original work, what makes the crypto art unique is the attached
technology used to double check and certify the transactions: each user can sell her own
crypto artwork only once and any attempt to sell the artwork twice is signaled out by
ep

blockchain technology as invalid. Few examples of very expensive crypto artworks are
“Everydays: the First 5000 Days” by Mike Winkelmann (alias Beeple) which is a collage
of 5,000 digital images created by the artist and has been sold for 69.3 million dollars by
Pr

Christie’s in its first NFT auction, and CryptoPunks 10,000 collectibles developed by
Larva Labs which are 10,000 low res images of unique digitally characters

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 algorithmically generated through computer code. With the most expensive of these

ed
collectibles exchanged between 2 and 7 million dollars each (see Figure 2).

To buy crypto art a prospective collector must first buy the currency of exchange
used for the transaction, which is always a cryptocurrency and in most cases is Ethereum

iew
(ETH), then the collector should open a digital wallet through the dedicated apps and
operate on the marketplace via its wallet. Each transaction the collector will initiate is
validated by blockchain technology and, just to give the broader picture, the energy
consumption required for validating each transaction is estimated around 48 kWh2.

v
This new framework to develop art content attracted more than just digitally

re
native unknown artists and tech-savvy collectors, the NFTs world seduced already
existing non-digital artists as well and provide fertile ground for new investment fund
with the aim of building wealth with NFTs trading that operates in still developing
er
marketplaces. This new environment is surely growing in popularity and economic
relevance.
pe
3. Methodology

The sample we use is made of secondary market transactions on Art and

Collectible tokens executed in the OpenSea marketplace and available through
Nonfungible.com. Considering the similarities from an asset pricing perspective, we treat
ot

Art and Collectible tokens as a single category: crypto art. The sample is made of 1,238
daily observations from January 2018 to August 2021. Figure 3 provides a graphical
tn

representation of the price for crypto art trades during the period. To test market
efficiency hypothesis for crypto art we follow Dowling (2021b) and rely on automatic
variance ratio (AVR) test from Kim (2009) and automatic portmanteau (AP) test by
rin

Escanciano and Lobato (2009). Additionally, we perform an autoregression model on

crypto art daily returns.
ep

2
See “What is cryptoart, how much does it cost and can you hang it on your wall?”, The Guardian,
Pr

March 4, 2021 (https://www.theguardian.com/culture/2021/mar/05/what-is-cryptoart-how-much-does-it-

cost-and-can-you-hang-it-on-your-wall).

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 The second part of our analysis refers to the risk factors driving the price of crypto

ed
art and the similarity, if any, between crypto artwork price and real-world artwork price.
In this analysis we are interested in understanding the underlying common pattern of risk
factors, rather than market microstructure features and frictions. Hence, we rely on a

iew
lower sampling frequency: we rely on weekly average prices, instead of daily data,
gathering 188 weekly observations. We first regress crypto art returns on several
independent variables such as cryptocurrency exchange rate returns, and stock index
returns. Then we perform the same empirical analysis on real world art index returns

v
using the stock index return as independent variable. As representative of prices for the
real-world art we consider different sources. We use Art Price 100 Index with 94 quarterly

re
observations from early 1998, and Sotheby’s Mei Moses Index with 234 monthly
observations also from early 1998. The Art Price Index is developed by the listed
company ArtMarket.com and available from Artprice.com, the Index is calculated based
er
on all Fine Art auction results (paintings, sculptures, drawings, photographs, prints,
watercolors, etc.) recorded by Artprice.com, apart from antiques and furniture. Different
pe
Sub-Indices are available according to the kind of art and the geographical area we want
to track. The Sotheby’s Mei Moses Index developed by Mei and Moses (2002) is
computed using over 80,000 repeat auction sales for the same object over time. Given
the repeated sales approach this second Index is, to the best our knowledge, the art
ot

index more comparable with financial markets.

The following statement presents our testable hypothesis.

tn

Hypothesis 1 (H1): Crypto art price does not behave efficiently and
is more related to the cryptocurrency in which its price is
denominated rather than real-world art with which crypto should
rin

share common features from an asset pricing perspective.

Our hypothesis is motivated by the idea that the market for crypto art is still
developing and - in absence of a sound regulation and a stable flow of money - secondary
ep

prices are influenced by trends in exogenous variables (i.e., cryptocurrencies exchange

rates with USD) that should be unrelated to art asset. This dependance from
cryptocurrencies weakens the understanding that crypto art is an artistic asset and
Pr

strengthen the idea that is, in the end, a storing facility for inflows of cryptocurrency

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 money, where cryptocurrencies are already perceived more as a speculative asset rather

ed
than an actual medium of exchange (see Baur et al. (2018)).

Equation 1 presents the econometric model used to test both our hypotheses:

iew
𝑐𝑟𝑦𝑝𝑡𝑜𝑎𝑟𝑡 = 𝛼 + 𝛼 𝑦 + 𝜀 (1)

with 𝑦 a vector of exogenous variables such as Ethereum (ETH) and Bitcoin (BTC) rate

v
of exchange with USD, traded volume in ETH, S&P 500 stock index (Equity Index), and

re
Crypto Market 10 Index (Crypto Index) provided by the SIX Swiss Stock Exchange and
tracks the behavior of the major cryptocurrencies. This model is then compared to
different real-world art indices return, where the variable 𝑦 is the Equity Index as
er
representative of the real economy.

4. Empirical analysis
pe
Table 2 shows descriptive statistics for crypto art returns and other selected variables.
As expected, crypto art delivers higher returns and volatility than Equity Index.
Surprisingly the artistic token is also riskier and more remunerative than
cryptocurrencies. Results are confirmed if we exclude from the analysis the first part of
ot

2021 when we observe a strong increase in prices for crypto art, as shown in Figure 3. As
transaction volume increases, the return on crypto art is higher. NFTs are also strongly
tn

correlated with its pricing currency, ETH, and other cryptocurrency-related variables,
BTC and Crypto Index. NFTs do not exhibit strong correlation with Equity Index (see
Table 3).
rin

4.1. Market efficiency

Table 4 presents our results for market efficiency analysis on crypto art returns. Both
ep

AVR and AP show a strong significance in support of the inefficiency of crypto art market
throughout the entire sample period. Our interest is also to understand if the market
efficiency for crypto art shows any sign of improvement during the during the selected
Pr

time interval. From Dowling (2021b) we understand that market is inefficient for virtual
real estate NFTs, yet improving in its pricing discovery component. Figure 4 provides the

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 AP test on one-year rolling basis for crypto art returns in comparison to Equity Index

ed
return. Market for crypto art is proven inefficient also on a rolling basis with just a minor
improvement in the last part of the period. Although trading volumes for crypto art are
recently, since 2021, more significant than in previous years, the same also occurs for

iew
volatility (see Figure 3). Hence the improvement in efficiency might be a false positive
result, also considering the lack of regulation in place in the NFTs market.

Lastly, we provide additional evidence supporting market inefficiency from the

autoregression analysis which shows a significant negative relationship with past returns

v
up to one week of trading (see Table 3).

re
4.2. Crypto art vs. real art

Table 5 presents a comparison between yearly return and volatility statistics for
crypto art and real-world art. Crypto art shows extreme returns equal to 219 percent on
er
yearly basis with an impressive volatility level equal to 318 percent3. The asset with
which crypto art correlates more is ETH, and this seems to be a less-risky alternative for
pe
the digital artworks with 112 percent mean return and 200 percent volatility. Equity Index
(S&P 500) shows results far from the two crypto-assets, from January 2018 to mid-2021
the stock market provided an average yearly return of 28 percent with 36 percent yearly
volatility. When we consider the world of real art the Art index global provides an average
ot

yearly return of approximately 1.4 percent and the Art index for contemporary art, which
can be more closely related to NFTs artwork, a return of 7 percent, with both of them
tn

showing an annual volatility little less than 30 percent during the last twenty years.
Sotheby’s Mei Moses Index offers an understanding of art prices when new art piece are
excluded from the analysis and only recursive trades on the same artworks are considered,
rin

within last twenty years the index provided an average yearly rate of return of barely 1
percent and less than 2 percent volatility.

In Table 6 we show the results from our model in Equation (1) and provide
ep

additional analysis on the driving features of crypto art returns and how this asset is
Pr

3
If we exclude from the sample the year 2021 where crypto art shows a strong increase in price (see
Figure 3) the average yearly return is 109 percent, and the volatility is 300 percent.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 related with other variables from the same fintech ecosystem. As expected from the

ed
correlation matrix reported in Table 2, Table 6 also shows that crypto art return are
strongly related to the exchange rate for its main currency of denomination, ETH, as well
as with the main cryptocurrency in circulation, BTC and the crypto currency index. More

iew
than from the former three, the digital asset is severely driven by the rise in its volume,
testifying that crypto art market is mostly a sellers’ market, as more traded Ethereum
drive crypto art return up. The strong positive relationship between crypto art prices and
volume demonstrates that crypto art market environment is still developing and with a

v
yet to be completed long-run market equilibrium, and only a minor similarity to emerging
art market can be found according to Kraeussl and Logher (2010).

re
As expected, crypto returns share no relationship with the Equity Index, testifying no
link with other financial assets and the economic factors that affect them. Crypto art
valuation is solely driven by the demand for the asset and the appreciation its
er
denominating currency. More than that, the demand itself is expressed in Ethereum, hence
we can easily state that the only driving force is the demand for Ethereum-denominated
pe
asset.

When we consider real-world art we assume that Equity Index plays, for this asset,
the same role that ETH, BTC and Crypto Index play for crypto art. Equity Index
represents the underlying economy to which real-world art ultimately belongs, and we
ot

assume there is a relationship between the two we should observe significant evidence
from our analysis, like we observe for crypto art and cryptocurrencies. However, we
tn

perceive a different behavior from the real-world art, all but the Art index based only on
US transactions share no relationship with the Equity Index, and even for Art index USA
the relationship is not as strong as the one we observe between crypto art and ETH. Given
rin

the extended time period of Art index and the quarterly frequency of sampling, we cannot
exclude that the two series share a common trend due to US inflation rate, when we adjust
the model for US inflation the relationship between Art index USA and Equity Index lose
ep

most of its significance with the overall explanatory power of the model increasing.
Finally, we consider Sotheby’s Mei Moses Index where only recursive trades are
considered, and the sampling frequency is higher, for this index we observe no
Pr

relationship with the underlying economy.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 5. Conclusions

ed
In this paper we examine the new and thriving world of NFTs artworks, where
crypto-money is traded for digital files representing images and videos with embedded a
certificate of guarantee built on blockchain technology. The market for crypto artworks

iew
is growing in trading volume and popularity. Hence, an understanding of its functioning
is needed. We follow Dowling (2021b) and find that the crypto art market is far from
being an efficient market as prices present strong autoregressive patterns lasting for
several days. We then compare the returns from crypto art with the returns from real-

v
world art and find that the former is severely more volatile than any other asset in our

re
sample and is strongly related to the increase in prices we observe in ETH and other
cryptocurrencies. In contrast, real-world art provides low returns and volatility and shares
no common ground with the stock market of the underlying economy.
er
We conclude that crypto art is far from being just another medium to express
artistic content. Demand for crypto art does not stem from demand for artworks by
collectors. On the contrary, given its uniqueness and its currency of denomination, it is
pe
perceived as haven for investors willing to invest in ETH and other cryptocurrencies. The
current environment does not provide an extensive menu of assets if someone intends to
invest in cryptocurrencies. Given the popularity of NFTs, crypto art seems to be the
prominent place to park crypto money. However, the current condition might evolve if
ot

private and public institutions start to issue equity or debt denominated in cryptocurrency,
triggering a decrease in demand for crypto art and a related reduction in prices. Probably
tn

at that point, the remaining demand for the asset will be related to the artistic content and
crypto art will belong to the same asset class as real-world artworks.
rin
ep
Pr

10

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 References

ed
Baur D. G., Hong K. H. and A. D. Lee, 2018. Bitcoin: Medium of exchange or speculative
assets?. Journal of International Financial Markets, Institutions and Money, 54,
177-189.

iew
Candela G. and A. E. Scorcu, 1997. A Price Index for Art Market Auctions. Journal of
Cultural Economics, 21, 175-196.

Dowling M. M., 2021a. Is Non-Fungible Token Pricing driven by cryptocurrencies?.

v
Finance Research Letters, forthcoming.

re
Dowling M. M., 2021b. Fertile LAND: Pricing Non-Fungible Tokens. Finance Research
Letters, forthcoming.

Escanciano, J.C. and I. N. Lobato, 2009. An Automatic Portmanteau Test for Serial
er
Correlation. Journal of Econometrics, 151(2), 140–149.

Higgs H. and A. C. Worthington, 2004. Art as an Investment: Risk, Returns, and Portfolio
Diversification in major Painting Markets. Accounting & Finance, 44(2), 257-
pe
271.

Kim, J.H., 2009. Automatic Variance Ratio Test under Conditional Heteroskedasticity.
Financial Research Letters, 6 (3), 179–185.
ot

Kraeussl R. and R. Logher, 2010. Emerging Art Markets. Emerging Markets Review,
11(4), 301-318.
tn

Mei J. and M. Moses, 2002. Art as an Investment and the Underperformance of

Masterpieces. American Economic Review, 92(5), 1656-1668.

Renneboog L. and C. Spaenjers, 2012. Buying Beauty: On Prices and Returns in the Art
rin

Market. Management Science, 59(1), 36-53.

ep
Pr

11

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Figure 1 – Growth in NFTs trading activity

ed
The Figure shows the time series of monthly volume in USD million for the entire NFT market and the
weight of the categories Art and Collectibles from early 2019 to mid-August 2021. Crypto art weight of
overall transactions is the sum of volumes for Art and Collectible categories available in the database.
Source: Nonfungible.com.

iew
700 100%

90%
600
80%

500 70%

v
60%
400
50%

re
300
40%

200 30%

20%
100
10%

0
er 0%
01-2019
02-2019
03-2019
04-2019
05-2019
06-2019
07-2019
08-2019
09-2019
10-2019
11-2019
12-2019
01-2020
02-2020
03-2020
04-2020
05-2020
06-2020
07-2020
08-2020
09-2020
10-2020
11-2020
12-2020
01-2021
02-2021
03-2021
04-2021
05-2021
06-2021
07-2021
pe
Volume in USD (mln) Crypto art (%)
ot
tn
rin
ep
Pr

12

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Figure 2 - Most expensive crypto art examples

ed
The Figure presents two of the most expensive crypto artworks. Sub. Figure (A) is “Everydays: the First
5000 Days” by Mike Winkelmann (alias Beeple) which is a collage of 5,000 digital images created by the
artist. The artowork has been sold for 69.3 million dollars by Christie’s. Sub Figure (B) presents the most
expensive CryptoPunks collectibles developed by Larva Labs with their price tag in ETH and USD.

iew
(A) – Everyday: the First 5000 Days by Beeple

v
re
er
pe
ot
tn

(B) – CryptoPunks collectibles largest sales

rin
ep
Pr

13

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 ed
Figure 3 – Crypto art weekly price

The Figure shows the time series of the average weekly crypto art price observed in the secondary market
from early 2019 to mid-August 2021.

v iew
re
er
pe
ot
tn
rin
ep
Pr

14

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Figure 4 – Portmanteau test for market efficiency

ed
The Figure shows the Portmanteau test for market efficiency for log-return on S&P 500 and Crypto art.

v iew
re
er
pe
ot
tn
rin
ep
Pr

15

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Table 1 – Variables description

ed
The Table describes the variables employed in our study for our testable hypothesis in Equation (1). The
sample is composed by 1,238 (188) daily (weekly) observations for Crypto art from January 2018 to mid-
2021, 94 quarterly observations for Art index and 234 monthly observations for Sotheby’s Mei Moses
Index from early 1998 to the end of 2020.

iew
Variable Description

Daily (weekly) log-return from the database provided by Nonfungible.com Crypto art
Crypto art represent the sum of all daily transactions from ‘Art’ and ‘Collectible’ categories in the
original database.

v
Average daily (weekly) log-difference in volume in Ethereum for crypto art
Volume

re
transactions.
Daily (weekly) log-returns for Ethereum to USD currency exchange (ETH) and Bitcoin
ETH, BTC
to USD.
Crypto Market 10 Index daily (weekly) log-returns. The Index is provided by SIX Swiss
Crypto Index Exchange and represents a weighted average of the performance of top
cryptocurrencies.
er
Equity Index S&P 500 daily (weekly) log-returns.

Risk-free rate 10 year US government bond yield rate.

pe
Quarterly log-returns on Art Price 100 Index. The index is developed by the listed
company ArtMarket.com and available from Artprice.com, is calculated based on all
Fine Art auction results (paintings, sculptures, drawings, photographs, prints,
watercolors, etc.) recorded by Artprice.com. Art Price 100 Index is composed by
different Sub-Indices based on artworks category and geographical region. In our
Art Price 100 Index
analysis we consider the Art index global that summarize all categories and
ot

geographical region, the Art index contemporary that relates only to contemporary art,
the Art index global euro-denominated to control for USD currency exchange and the
Art index USA which relates only to the US market. All Art indices but the euro-
denominated index, are USD-denominated.
tn

Monthly log-returns on Sotheby’s Mei Moses Index (Mei and Moses (2002)) uses the
Sotheby’s Mei purchase prices of the same painting at two distinct moments in time (i.e., repeat-sales)
to measure the change in the value of unique works of art. Based on approximately
Moses Index 80,000 repeat sales from 1810-present, Sotheby’s Mei Moses The methodology is based
on the Case-Shiller Real Estate Index.
rin
ep
Pr

16

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 ed
Table 2 - Descriptive statistics

The Table provides descriptive statistics for daily and weekly returns for the selected variables of study.

iew
Daily returns

Variable N. Mean SD Min Max Median Skewness Kurtosis

Crypto art 1,238 0.0052 0.6081 -2.1028 2.8588 0.0056 0.2003 4.4417

Volume 1,238 1.0396 4.6122 0.0006 70.5525 0.0093 8.8995 101.2212

ETH 1,238

v
-0.0003 0.0579 -0.5756 0.2134 0.0002 -1.6184 17.5809

BTC 1,238 0.0007 0.0455 -0.4940 0.1694 0.0019 -1.7622 22.2920

re
Crypto Index 1,238 -0.0007 0.0684 -0.7150 0.7473 0.0000 -0.5417 51.0270

Equity Index 1,238 0.0006 0.0129 -0.0999 0.0897 0.0008 -0.1416 16.3563

Risk-free rate 1,238 2.1720 0.7264

er 1.1155 3.1275 2.4090 -0.2473 1.3173

Weekly returns

Variable N. Mean SD Min Max Median Skewness Kurtosis

pe
Crypto art 188 0.0457 0.4601 -1.6335 1.4130 0.0526 -0.4836 4.5715

Volume 188 12.5153 57.2009 0.0110 545.4499 0.0734 7.0405 57.0234

ETH 188 0.0233 0.2882 -0.5475 3.4616 0.0049 9.0369 109.2059

ot

BTC 188 0.0187 0.2122 -0.3927 2.5077 0.0007 8.5936 102.0899

Crypto Index 188 0.0179 0.2148 -0.3937 2.5485 0.0024 8.7477 104.3776

Equity Index 188 0.0058 0.0517 -0.1307 0.6267 0.0065 9.0872 112.0831
tn

Risk-free rate 188 2.1691 0.7282 1.1400 3.1025 2.3970 -0.2366 1.3096
rin
ep
Pr

17

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Table 3 - Pairwise correlation

ed
The Table provides pairwise correlations for the selected variables.

Crypto
Variables Crypto art Volume ETH BTC Equity Index
Index

iew
Crypto art 1.000
Volume 0.818*** 1.000
ETH 0.374*** 0.291*** 1.000
BTC 0.308*** 0.250*** 0.946*** 1.000
Crypto Index 0.357*** 0.285*** 0.974*** 0.987*** 1.000
Equity Index 0.164* 0.116 0.840*** 0.802*** 0.820*** 1.000

v
Risk-free 0.153* 0.143* 0.096 0.123* 0.113 0.264***
*** p<0.01, ** p<0.05, * p<0.1

re
er
pe
ot
tn
rin
ep
Pr

18

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Table 4 - Market efficiency tests for crypto art

ed
The Table presents the market efficiency tests we performed on our dataset for crypto art returns. AVR is the
automatic variance ratio test from Kim (2009), and AP is the automatic portmanteau test by Escanciano and
Lobato (2009). AR1-10 are four autoregressive model with lags from one to ten. The Table presents, for AVR
and AP report the p-value of the test and for the autoregressive models both the coefficient and the robust

iew
standard error.

Lag AVR AP AR1 AR2 AR5 AR10

Automated 0.0000 0.0000

-0.389*** -0.495*** -0.584*** -0.597***
1

v
(0.030) (0.030) (0.030) (0.031)
-0.272*** -0.437*** -0.468***
2
(0.030) (0.036) (0.037)

re
-0.293*** -0.338***
3
(0.035) (0.037)
-0.219*** -0.286***
4
(0.034) (0.038)
-0.093*** -0.184***
5
(0.030) (0.038)
6
er -0.142***
(0.039)
-0.068*
7
(0.036)
pe
-0.072**
8
(0.037)
-0.053
9
(0.033)
-0.025
10
(0.030)
Observations 1,238 1,238 1,238 1,238 1,238 1,238
ot

Wald χ2 167.14 284.42 377.25 406.99

tn
rin
ep
Pr

19

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Table 5 - Crypto art and real-world art yearly descriptive statistics

This Table provides descriptive statistics for yearly return and standard deviation for crypto art, selected

ed
exogenous we use in Equation (1) model and the real art index. Crypto art is

Crypto Equity Art index Art index Sotheby’s Mei

Measure ETH
art Index global contemprorary Moses Index

iew
Mean return 2.1936 1.1184 0.2784 0.0144 0.0701 0.0084

Volatility 3.1877 1.9967 0.3582 0.2918 0.2805 0.0183

v
re
er
pe
ot
tn
rin
ep
Pr

20

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 Table 6 - Regression analysis on crypto art return vs real-world art return

ed
w
The Table provides the regression analysis for the entire sample when crypto art is the dependent variable. The sample is composed of 188 weekly observations for Crypto art
from January 2018 to mid-2021, 94 quarterly observations for Art index and 234 monthly observations for Sotheby’s Mei Moses Index from early 1998 to the end of 2020.

Variable
(1) (2)
Crypto art

(3) (4) (5) (6)

Art index
global

(7)
Art index
contemporary

(8)
Art index
global EUR

(9)

v i eArt index
USA

(10)
Art index
USA (ex infl.)

(11)
Sotheby’s Mei
Moses index

(12)

ETH
1.282***
(4.385)
1.167***
(7.915)

r e
r
0.693*** 0.706***
Volume
(14.202) (12.842)

e
1.355***
BTC
(3.621)

e
1.593***
Crypto Index
(3.947)

p
3.121 0.124 0.294 0.140 0.092** 0.0803* 0.009
Equity Index
(1.648) (0.801) (1.283) (0.974) (2.067) (1.750) (1.171)

t
0.031 -0.005 0.021 0.035 0.032 0.046 0.002 0.013 0.001 0.002 -0.0015 0.001
Constant
(0.813) (-0.235) (0.845) (0.918) (0.845) (1.142) (0.110) (0.875) (0.044) (0.452) (-0.219) (1.902)

o
Observations 188 188 188 188 188 188 94 94 94 94 94 234
R-squared 0.140 0.743 0.627 0.095 0.127 0.027 0.006 0.038 0.009 0.042 0.052 0.008
Std. Errors Robust
t-values are in parentheses
*** p<.01, ** p<.05, * p<.1
Robust Robust

t
Robust

n Robust Robust Robust Robust Robust Robust Robust Robust

r in
e p
P r 21

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645
 ed
v iew
re
er
pe
ot
tn
rin
ep
Pr

22

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4200645