The Construction of an Investment Risk Measurement Model and

Risk Avoidance Strategy under the Financialization Mode of Chinese

Artworks

SHANGJIN XIE*, QUANLIN LI

International College,

Krirk University,

Bangkok 10220,

THAILAND

*Corresponding Author

Abstract: - Along with the continuous development of artwork in the investment market, as a special investor

asset, it plays an important role in dispersing the investment risk and stabilizing the investment income for the

investors, and therefore it has become an important way of investor's asset portfolio allocation. The

financialization mode of Chinese artwork has become an important research direction in recent years, and the

investment risk under this mode is also a key factor that needs to be seriously considered. Based on this, this paper

explores the analysis and prediction of the results such as the return of China's art index based on the GARCH

model, and further explores the key factors affecting its risk, accordingly. It is verified that the GARCH model at

the 5% confidence level plays a role in predicting the maximum loss of investing in Chinese artworks. At the

same time, the corresponding optimization suggestions are put forward in terms of choosing formal purchasing

channels, learning relevant appraisal knowledge, improving relevant laws, disclosing information on art

securitization transactions, and using securitization and other channels. To be able to provide certain references

for the construction of the investment risk metrics model and the avoidance of the risk under the mode of China's

financialization of artworks.

Key-Words: - Art market, Art financialization, Investment risk, Metric model, Risk avoidance, Works of art,

investment yields, price indices.

Received: January 23, 2024. Revised: July 6, 2024. Accepted: August 4, 2024. Published: September 4, 2024.

1 Introduction

With the deepening financialization of artworks, the

risk of investment is also increasing. Macro-wise, the

economic downturn and the start of the bull market

are some of the reasons for the shrinkage of the art

auction market, [1]. In the short term, the volatility of

China's art trading market has increased, and the

investment risk has increased. How to effectively

measure the risk of art investment returns and use

relevant models to make predictions has an important

role for art investment. Based on the optimal

portfolio calculated from the 1875, S&P-2002

Treasury Index, the study shows that artwork can

optimize the traditional asset portfolio in the specific

proportions of 18.21%, stocks 27.69%, and bonds

54.10%, [2]. The results of a study on the returns of

artworks from different genres and countries using

the characteristic price method show that art is a

high-risk, high-return asset, [3]. In this context, this

paper takes China's art investment risk as an entry

point, selects the most representative branch of

China's art market - the painting auction market as a

data sample, combs its investment risk factors,

constructs China's art price index, establishes a

financial measurement model to calculate the yield

and volatility of art investment, and finally makes a

recommendation on the future construction of

China's art market.

2 The Construction of China's Art

Index

2.1 Data Source

This paper adopts the statistical data of the Art

Market Monitoring Center of Artron (AMMA) and

selects the Chinese painting 400 and oil painting 100

component indices from 2006 to 2017 to construct

China's artwork price index.

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

384

Volume 12, 2024

2.2 Index Construction

This paper draws on the method of constructing

indexes by Wang Shuo (2016) to construct China's

artwork index with the above two-component

indexes as data samples and turnover as weights, and

the expression is:

tttt YYGGt PPP





(1)

and

represent the National Painting 400

index and the Oil Painting 100 index in period t.

tt YandG

denote the turnover weights in the

corresponding periods.

3 Processing of Data and its Intrinsic

Characteristics

3.1 Data Processing

Quarterly data is used in this paper. Therefore, the

data are processed using the interpolation method to

fill in the missing quarterly data and convert the

frequency of the artwork index from semi-annual to

quarterly.

3.2 Intrinsic Characteristics of Chinese

Artwork Index Return Series

3.2.1 Descriptive Statistics and Normality Test

This paper uses the processed quarterly Chinese

artwork index data, with a data volume of 48. Next,

the quarterly return on Chinese artwork investment is

calculated, ignoring the transaction fees and taxes of

the artwork market in the calculation process, and the

formula is as follows:

 

/log 

ttt PPR

(2)

The quarterly rate of return of China's art index

for the period of 2003-2014 will be made icons as

follows Figure 1:

Fig. 1: Time series graph of quarterly return of China

Art Index

From the time series graph of quarterly returns

of the China Art Index in Figure 1, the quarterly

returns of artworks reflect strong volatility. Therefore,

for the reliability of the conclusions of the data

analysis, the following test of data normality is

conducted, and this paper uses the most widely used

Q-Q (quantile-quantile) test, [4]. Figure 2 shows the

histogram and descriptive statistics of the distribution

of Chinese art index returns and the Q-Q plot of the

normality test.

Fig. 2: Q-Q plot of the histogram of the distribution

of returns of Chinese art index and descriptive

statistics and normality test

The characteristic statistics and normality test

results of artwork investment returns are given in

Figure 2. From the results of descriptive statistics,

from the mean return (Mean), the quarterly average

return is 5.03%, relatively at a high level; from the

standard deviation (StandardDeviation), the return

fluctuates 11.79%, obviously with a large volatility,

indicating that there is a considerable gain and loss in

the investment of artworks, and its risk cannot be

ignored; from the skewness (Skewness), the return

fluctuates 11.79%, obviously with large volatility,

indicating that there is a considerable gain and loss in

the investment of artworks, and its risk cannot be In

terms of Skewness and Kurtosis, the skewness

coefficient of the return series is less than 0, and the

kurtosis coefficient is slightly less than 3, which

indicates that the series is generally consistent with

the standard normal distribution, and the Jarque-Bera

statistic does not reject the assumption that the time

series is normally distributed at the 95% confidence

level, which suggests that normal distribution is

suitable for modeling the changes in quarterly returns

of Chinese artworks, [5].

3.2.2 Smoothness Test

The GACRH model is only suitable for modeling

smooth series, so the time series of yields are tested

for unit root (Augmented Dichkey-Fuller Test, ADF),

and the results of the ADF test are shown in Table 1.

Table 1. ADF unit root test results

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

385

Volume 12, 2024

T-statistic

Prob.

ADF

statistical

values

-4.168423

0.0017

confidenc

e level

-3.550396

-2.913549

10%

-2.594521

The statistical value of ADF obtained through

the unit root test is -4.168423, the absolute value is

greater than the critical value at different confidence

levels, which can indicate that the time series of

China artwork index return is smooth, [6].

3.2.3 Serial Correlation Test

Before establishing the model, the correlation and

autocorrelation of the time series data should also be

examined, according to the autocorrelation function

(Auto-Correlation Function (ACF)) and partial

autocorrelation function (Partial-Auto-Correlation

Function (PACF)) of the time series of the return on

Chinese art index tR. The autocorrelation coefficient

of the time series

as well as the partial

autocorrelation coefficient are significantly non-zero,

and the Q statistic is also significant, [7].

Fig. 3: Autocorrelation and bias correlation test plots

of the return series of China's artwork index

By analyzing the correlation and partial

correlation plots in Figure 3, it is known that the time

series of quarterly artwork returns is smooth. The

first-order autocorrelation coefficient is 0.552, the

second-order autocorrelation coefficient is 0.142, but

from the third order, the autocorrelation coefficients

are close to 0. From the p-value, it can be seen that

the autocorrelation coefficients and the bias

correlation coefficients fall within the confidence

interval and are statistically significant, [8]. Then the

AR (2) model is established for research, and it is

found that the residuals cannot satisfy the assumption

of normality. The time series plot of the residuals is

made as in Figure 4 and it is found that there is a

volatility agglomeration in the time series of the

artwork yield, and the variance of the residuals has

the characteristic of changing over time, which

shows that there is a phenomenon of conditional

heteroskedasticity in the sequence of the error term.

Fig. 4: Residuals series

3.3 Model Setting

3.3.1 Introduction of VaR methods

The VaR method is one of the most common and

important financial risk measurement tools with the

widest range of applications among many financial

risk measurement methods, [9]. The standard method

commonly used in modern financial risk

measurement is the VaR (Valueat Risk) method,

based on the characteristics of the VaR method which

is concise and easy to understand, this paper selects

the VaR method to study the investment risk of

artwork. VaR (ValueatRisk) is essentially a

quantitative study of the value of asset value

fluctuations, and a key step in the process is to

construct a probability distribution of the change in

the value of the asset. probability distribution. [10].

3.3.2 Calculation of VaR based on GARCH

Modeling

(1) Conditional Heteroskedasticity and GARCH

Modeling

Some foreign scholars have studied the effectiveness

of the art market from an empirical point of view, and

their results show that there is autocorrelation in art

returns. The most prominent feature of the ARCH

model is that it gives the method of calculating the

conditional variance of the time series, i.e., the

conditional variance of the ARCH process can be

derived by constructing a function of all kinds of

stochastic disturbances in the previous time at each

moment t. Therefore, the ARCH model can describe

the volatility set because of external disturbances in a

better way. Volatility agglomeration due to external

disturbances. The model form of GARCH (p, q) is as

follows:

 









jjtj

iitit

ttt





(3)

Where



is a vector of explanatory variables that

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

386

Volume 12, 2024

have an effect on

, and the residual return term



obeys mean

 

1

tt



with variance

 

1hE tt 





. For the GARCH model to be

meaningful, the following conditions must also be

satisfied:

1 1













  

 

jji



(4)

(2) ARCH effect test for serial residuals

The purpose of the ARCH effect test is to test

whether the model describes the heteroskedasticity of

the return time series well. It is common in academia

to use the Lagrange multiplier method (ARCH-LM

test) to test whether the random disturbance term Ԑt

in the model has an ARCH effect. Create an auxiliary

regression equation:

110 ... qtqtt

h 



(5)

Test for the presence of ARCH effects in the

series, i.e., the regression coefficients



,,...,,121 

in the above equation at least one

is not zero, the original and alternative hypotheses of

the test are:

 

qiH





10:

0...:

210





(6)

Test statistic:

 

qnRLM 22 ~





(7)

Given the significance level



and the degree of

freedom

, if

 

qLM 2





, then

is rejected

and there is a

-order ARCH effect in the series; if

 

qLM 2





, then

cannot be rejected, which

indicates that the series does not have a

-order

ARCH effect.

3.4 Tests and Analysis of Regression Results

3.4.1 Regression results

After the intrinsic characterization of the time series

of China's artwork investment returns in the previous

section, based on the results of the test results, the

first-order lag

1t

and second-order lag

2t

the independent variables of the regression equation,

and tR as the explanatory variables. Regressions

were done on the regression equations with different

orders of

and

conditional variances, and the

most appropriate orders of

and

in the

GARCH model were selected based on the Lagrange

multiplier test for the significance of the LM statistic

and the values of AIC (Akaike Information Criterion)

and SC (Schwarz Criterion) for the different orders of

the model. The more significant the LM statistic is,

the greater the possibility of ARCH effect in the

residuals; at the same time, the predictive ability of

the model changes with the values of AIC and SC, if

the values of AIC and SC are smaller, it means that

the selection of lag order of the variables in the model

is more appropriate, and the predictive ability is more

powerful (Table 2).

Table 2. AIC and SC results for different

GARCH (p,q) models

AIC

GARCH (1,1)

-2.034953

-1.801052

GARCH (2,1)

-1.808299

-1.535416

GARCH (1,2)

-1.912437

-1.639554

GARCH (2,2)

-1.773409

-1.461542

Note 1: The values of AIC and SC increase when the order

increases after

and

. The specific values are not listed here.

In this way the GARCH (1, 1) model was finally

chosen:

22110







ttt

tttt

RcRccR





(8)

The logarithmic yield series ｛tR｝ is estimated

using the great likelihood method. The final GARCH

model is:

815542.0104199.0015347.0

202251.0756436.0018917.0







ttt

tttt

RRR



(9)

The regression uses a total of 48 data, and the

is 0.3618, indicating that the model has a strong

explanatory ability and better fits the data changes.

From the results of the regression, the coefficients



and



are highly significant and satisfy the

condition

 

10 11 



, indicating that the

model is meaningful. From the coefficients of the

lagged one and two periods, it can be seen that the

artwork yield fluctuates dramatically and is highly

influenced by the previous period.

3.4.2 Result Analysis and Prediction of GARCH

Model

(1) Result analysis

In this paper, concerning a large number of domestic

research literature on GARCH model, the following

four most commonly used indicators to measure the

forecasting effect of time series are selected: ①The

square root of the average forecast error sum of

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

387

Volume 12, 2024

squares (Root-Mean-SquareError (RMSE)), which

indicates the degree of deviation of the forecasted

value from the true value, and the smaller the RMSE

the more accurate the result is ②Theil Inequality

Coefficient (TIC), TIC between 0 and 1, when TIC =

0, it means that the predicted value and the actual

value are in a state of the perfect fit, at this time the

model has the strongest predictive ability; TIC = 1 is

the worst predictive ability of the model ③Mean

Absolute Error (MAE) ④Mean Absolute

Proportional Error (MAPE) (Table 3).

Table 3. GARCH model returns prediction error

results

Norm

RMSE

TIC

MAE

MAPE

GARCHpredicted

values

0.093214

0.428633

0.065717

79.4164

Fig. 5: Comparison of China Art Index Returns and

GARCH Model Predicted Returns

From the prediction results (Figure 5), the

predicted art investment return rate carried out by the

GARCH model tends to be consistent with the actual

art investment real value chart, with a good fit, and

the four prediction errors of RMSE, TIC, MAE, and

MAPE are within a reasonable range. The test results

prove that the GARCH (1, 1) model applies to the

time series of China's quarterly artwork returns, and

can predict the short-term artwork investment returns

more accurately within a certain error range.

(2) VaR predictive value

According to the above autoregressive conditional

heteroskedasticity GARCH (1,1) model calculation

results, the conditional variance th over time will be

substituted into the formula

ˆhZrVaRt





where



is the upper quartile of the standard normal

distribution, without loss of generality, at the

confidence level of 5%, that is

05.0



, the

calculation of the artwork investment yield VaR

value at time t. The calculated quarterly predicted

VaR values for the artwork index are plotted on a

graph with the corresponding actual values. As seen

in Figure 6, the quarterly VaR curve calculated

according to the model results is consistent with the

trend of the real investment return curve, indicating

that the GARCH model plays a role in predicting the

maximum loss of investing in China's artworks at the

5% confidence level.

Fig. 6: Comparison of China Art Index Returns and

VaR Values

This chapter uses financial risk measures and

applies the GARCH model to model the value-at-risk

(Va R) of art market returns in China, and empirically

examines the volatility of art index returns using a

sample of data from China's art auction market. It is

found that (1) China's artwork quarterly returns have

the characteristics of a general financial time series

and are consistent with the assumption of normal

distribution, which meets the conditions for risk

measurement using the GARCH-VaR type model (2)

China's artwork index return time series have strong

autocorrelation, and measuring the risk of artwork

investment returns can be done using the Va R

method based on the GARCH model, which can

better characterize the distribution of the art market

yield series, make reasonable predictions and truly

reflect the risk situation of the art market.

4 Art Market Investment Risks

4.1 Preservation Risk

The uniqueness of the artwork and the fixed nature of

the product form and other characteristics of the

artwork preservation have brought great difficulties.

As we all know, the character of the artwork

determines the preciousness of the artwork to a large

extent, and most collectors are obsessed with

pursuing the integrity of the artwork, which is also

the reason why a large number of fakes are produced.

However, those with historical and cultural

precipitation of art, after years of baptism, can be

preserved intact, and how many? Most of the

collectors understand this truth - the storage of works

of art is not an easy task, [11]. Of course, it is not

possible to generalize, different works of art will

have different impacts if they are subjected to the

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

388

Volume 12, 2024

same degree of wear and tear.

4.2 Risk of Authenticity

With the continuous development of the art

investment market, the emergence of a large number

of business opportunities at the same time also brings

greater risk to the art buyer --- the authenticity of the

artwork to identify. This is the most important point

in the art investment risk. The asymmetry of

information in auction houses often causes great

difficulties for buyers, and many auction houses that

do not operate in a standardized manner deceive

buyers through this, taking the opportunity to make a

lot of profit, and even forgeries can be sold at a pretty

good price.

4.3 Market Risk

The art market, like the financial market, is divided

into primary and secondary markets. The primary

market consists of galleries, art suppliers, and art

buyers; the secondary market is through the auction

company and other media organizations, and will be

sold in the primary market of art again into the

market. The standardization of the art market is the

standardization of the art primary market and

secondary market participants, and our country

currently does not have a strong management system

in this regard. Auction houses, brokers, etc. in the

evaluation of the merits of a work of art at the same

time, but also to the buyer to bring a certain role in

guiding, but also may be the buyer to make the wrong

guidance, to the trading market to bring a great

credibility crisis.

4.4 Realization Risk

Art investment for profit is based on a higher selling

price than the purchase price. Internationally, the art

investment cycle is generally the shortest for 3 years,

as long as 10 years. Medium and long-term

investment makes art more difficult to realize, if art

holders in the holding period are in urgent need of

cash, many times will have to sell at a low price,

resulting in art investment to the investor to bring

losses. The preferences of different collectors will

also have a risk on the delivery of works of art, may

bear the pain in the auction house to buy down the art

simply cannot find the person who takes over.

5 Risk Avoidance Suggestions

5.1 Choose Formal Purchase Channels

By changing the purchase channel, investors can

effectively avoid the market risk and the risk of

authenticity. When choosing investment methods, we

should try to choose a good reputation in the industry,

which has become a large-scale company or auction

house. Formal institutions have a certain

identification ability, and can do for investors to art

quality control, [12].

5.2 Learning Relevant Identification

Knowledge, Increasing the Training of

Talents

Investors should have a certain ability to identify the

authenticity of works of art, in the purchase of

investment art to maintain sanity, to maximize the

avoidance of vicious price gouging, and due to the

negative evaluation of art brokers brought about by

the purchase of risk. Staying away from some of the

more valuable masterpieces and choosing

contemporary artworks that are easy to identify for

investment is also a wise way to buy. On the other

hand, each institution should increase the training of

art appraisal talents, organize regular training of

appraisal knowledge, optimize the art market

environment suitable for the development of talent

generation, and improve the credibility of the

institution.

5.3 Publicize the Information on Art

Securitization Transactions

The current serious information asymmetry between

domestic auction houses and art investment

institutions has brought serious obstacles to art

securitization transactions and great risks to investors.

Relevant policies should be formulated to disclose

information on artwork transactions, number of

transactions, sources of transactions, artwork

business conditions, information on associated

personnel, and assessment opinions of professional

institutions, etc., to provide a relatively equal

platform for securitization of artworks, and to make

artworks better organically integrated with the

financial market.

5.4 Through Securitization and other

Channels, Let Artworks Go to Mass

Consumption

Since artworks are generally expensive and have a

high investment threshold, they are often favored by

high-net-worth people. To effectively avoid the risk

of realizing art investment, only by the high net

worth crowd is not enough, should be art investment

expand to the public investment, the formation of a

large base population size of the investment market.

This requires on the one hand to improve the artistic

cultivation of the public, on the other hand, through

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

389

Volume 12, 2024

the securitization of artworks into multiple small

shares, lowering the threshold of public investment.

6 Conclusion

This paper constructs an investment risk

measurement model in the context of the

financialization of Chinese artworks. At the same

time, this paper further puts forward the main

influencing factors of art market risk and risk

avoidance strategies, to provide certain references for

the efficient implementation of risk measurement and

the long-term development of art financialization

mode in China. However, because the investment in

the art market integrates the influence of the

government's macro-control, market operation

mechanism, and other multiple factors, there are still

some imperfections in the research process. In future

research, we will study the risk of art market

investment more comprehensively and deeply from

multiple perspectives. Emphasis will be placed on

the emerging art market (such as free trade zones), art

investment model (network auction, art fund), and

other aspects of the study.

References:

[1] Zhao HJ, Cao ZY. Investment risk measurement

and assessment in countries along the belt and

road: An analysis based on the panel data of

principal component method. International

Business Research, Vol.44, No.02, 2023, pp.

14-28. DOI: 10.13680/j.cnki.ibr.2023.02.008.

[2] Campbell R, Hulsman R, Koedijk KG. Optimal

Portfolio Selection in a Value-at-risk

Framework. Journal of Banking and Finance,

Vol.34, No.09, 2014, pp. 1789-1840.

[3] Higgs H, Worthington A. Financial Returns and

Price Determinants in the Australian Art Market:

1973-2003. The Economics Record, The

Economic Society of Australia, Vol.03, No.81,

2015, pp. 113-123.

[4] Mu C. F. Art investment yield, monetary policy

and corporate risk taking. Finance and

accounting newsletter, Vol.01, No.22, 2015, pp.

69-74. DOI:

10.16144/j.cnki.issn1002-8072.2022.15.026.

[5] Liu H. Y. Discussion on the problems and

countermeasures of Chinese artwork

financialization. Business Culture, Vol.01,

No.04, 2022, pp. 33-35. DOI:

10.13939/j.cnki.zgsc.2022.35.048.

[6] Zhu J. Overview of domestic and international

research on financialization of art market.

Chinese Painting and Calligraphy, Vol.01,

No.04, 2021, pp.122-125.

[7] Zhang ZY, Liu HL. Progress of art finance

theory research. Economics Dynamics, Vol.18,

No.03, 2020, pp.1-13.

[8] Foteini V, Antonis B, Melissa T, Chris S,

Andrew H, Konstantinos C. Crypto Collectibles,

Museum Funding and OpenGLAM: Challenges,

Opportunities and the Potential of

Non-Fungible Tokens (NFTs). Appl.

Sci. 2021, 11(21), 9931, https://doi.org/

10.3390/app11219931 - 24 Oct 2021.

[9] Kuo Y. X. Optimization Analysis of Risk

Control and Regulatory System in the Era of Art

Financialization. China Market, Vol.24, No.02,

2022, pp. 54-58. DOI: 10.13939/

j.cnki.zgsc.2022.24.054.

[10] Xie C, Hu YB, Long JY. Research on market

risk and credit risk measurement of securities

and investment securities investment funds and

their relationship. Finance and economics

theory and practice, Vol. 40, No.03, 2019,

pp.52-58, DOI:

10.16339/j.cnki.hdxbcjb.2019.03.008.

[11] Fu Y, Zhang JZ, Zhou C. Research on P2P debt

investment strategy and risk measurement

issues based on mean-variance model.

Management Review, Vol. 29, No.07, 2017, pp.

19-28, DOI:

10.14120/j.cnki.cn11-5057/f.2017.06.008.

[12] Wang X, Cai JL, Tang L. Research on VaR risk

measure of stock portfolio based on

BEMD-Copula-GARCH model. Systems

Engineering Theory and Practice, Vol. 37,

No.02, 2017, pp.303-310.

Contribution of Individual Authors to the Creation

of a Scientific Article (Ghostwriting Policy)

- Shangjin Xie contributed to the study conception and

design.

- Quanlin Li carried out the material preparation, data

collection and analysis.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.

Conflict of Interest

The authors have no conflicts of interest to declare.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the Creative

Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en_

WSEAS TRANSACTIONS on COMPUTER RESEARCH

DOI: 10.37394/232018.2024.12.38

Shangjin Xie, Quanlin Li

E-ISSN: 2415-1521

390

Volume 12, 2024