The effect of liquidity and arbitrage on the price efficiency of Chinese ETFs

• Published date: 01 December 2023
• Author: Yuan Fu,Christine Jiang
• Date: 01 December 2023
• DOI: http://doi.org/10.1111/jfir.12349

Received: 25 December 2022

|

Accepted: 12 July 2023

DOI: 10.1111/jfir.12349

ORIGINAL ARTICLE

The effect of liquidity and arbitrage on the price

efficiency of Chinese ETFs

Yuan Fu|Christine Jiang

Department of Finance, Fudan University,

Shanghai, PR China

Correspondence

Christine Jiang, Department of Finance,

School of Management, Fudan University,

Shanghai 200433, PR China.

Email: christine_jiang@fudan.edu.cn

Abstract

We study the potential factors that determine the large and

persistent price deviations in Chinese equity exchange‐traded

funds (ETFs). Our results suggest that ETF liquidity and arbitr-

age activity are positively correlated with ETF price efficiency,

and the relation is more pronounced with higher institutional

ownership. We also evaluate the effect of two exogenous

shocks in the Chinese market. Using a policy change that

added market makers to ETFs on the Shenzhen Stock Ex-

change (SZSE) and Shanghai Stock Exchange (SSE), we find that

market makers improve price efficiency and that the impact is

stronger for ETFs with lower liquidity. We also exploit a change

in trading rules on the SZSE and show that the relaxation of

arbitrage restrictions improves price efficiency. Altogether,

these findings provide evidence that lack of liquidity, due to the

unique market structure and regulations of the Chinese market,

contributes to price inefficiency of Chinese ETFs.

JEL CLASSIFICATION

G12, G14

1|INTRODUCTION

The importance of financial innovation is widely recognized.

1

Exchange‐traded funds (ETFs), one of the most

important financial innovations in recent decades, have transformed the asset management industry by providing

investors with not only a more liquid and lower cost alternative to mutual funds but also access to previously

J Financ Res. 2023;46:1103–1140. wileyonlinelibrary.com/journal/JFIR

|

1103

© 2023 The Southern Finance Association and the Southwestern Finance Association.

1

Many leading scholars, including Miller (1986)andMerton(1992), highlight the importance of new products and services in the financial market and suggest that

the addition of a new type of security (e.g., futures and options) can complete the financial market (Duffie & Rahi, 1995), lead to more risk sharing (Allen &

Gale, 1994), and promote more hedging and informed trading (Dow, 1998), but also create complexity to exploit the users (Allen, 2012).

unavailable asset classes. The total assets under management (AUM) of ETFs in the United States has grown

six‐fold since 2009 to over $6 trillion in 2021, exceeding the size of the hedge fund industry.

2

The Chinese ETF

market has experienced similar rapid growth, with AUM growing from 150 billion yuan in 2013 to 1.2 trillion yuan

(approximately $200 billion) in June 2021.

3

The popularity of ETFs has attracted increasing attention from academic

researchers, market participants, and regulators to better understand ETFs as a financial innovation.

Although several studies have evaluated the market impact of ETFs,

4

the price efficiency of ETFs,

particularly in emerging markets, has received limited attention. In a well‐functioning market, ETF prices are

kept in line with the prices of their underlying assets, or net asset value (NAV), through arbitrage between

primary and secondary markets. However, ETFs are not always priced efficiently (Engle & Sarkar, 2006).

Inefficient ETF prices can distort market signals and disperse risk into the underlying assets through creation

and redemption activities that drive prices away from the fundamental value, increase volatility, andeven

promote market instability. Previous studies have linked the efficiency of ETF prices to trading costs and limits

to arbitrage, such as holding illiquid assets, market stress, or stale NAV prices (Madhavan, 2012;

Petajisto, 2017). Although price deviations are generally small for equity ETFs in the United States, this may

not be the case in emerging or other markets because of differences in market design and arbitr age restrictions.

Broman (2020) indicates that mispricing in the European ETF market is several times higher compared to that in

the US market because of the structure of multiple cross‐listings and higher arbitrage costs imposed through

settlement systems.

5

Given the paucity of studies on the price efficiency of ETFs in emerging markets such as

China, we first document distinct characteristics of the Chinese ETF market and then analyze the potential

factors that determine price inefficiencies of Chinese equity ETFs, including liquidity and relative liquidity,

arbitrage environment and activity, and ownership structure.

Substantial variations in trading processes and market structures exist between the Chinese and US markets.

Importantly, restrictions on short selling and T+1 trading make the well‐functioning arbitrage mechanism that ties

prices and NAVs closely together forUS ETFs not directly applicable for Chinese ETFs. Accordingly, we identifytwo

importantdifferences between the ChineseETF market and more developedmarkets such as the US market.First, we

provide evidencethat price deviations of ChineseETFs are substantially largerand more persistent than thosefor US

ETFs. The equal‐weighted (value‐weighted) mean value of the deviation is 0.586% (0.204%), as opposed to around

0.022% (0.037%)for US equity ETFs (Jain et al., 2021;Petajisto, 2017). Deviationscan also persist on average for5.53

days in China, approximately 10 times longer than in the US. Second, the liquidity of Chinese equity ETFs is mostly

lower than the liquidity of their underlying portfolios, suggesting that relative liquidity is negative. One of the most

well‐established advantages of ETFs in the United States and other developed markets is increased liquidity, or

positive relative liquidity.

6

We find that this advantage does not exist in the Chinese ETF market and is actually

reversed. These differences prompt us to investigate the causes of the significant and persistent price deviations in

Chinese ETFs: whether the unique restrictions on the Chinese market are to blame and how the interplay of the

liquidity of Chinese ETFs and that of their underlying portfolio affect the price efficiency of Chinese ETFs.

2

Data are collected and compiled from the ETF database (www.etfdb.com).

3

Data are collected and compiledfrom the Wind database.

4

Some believe that ETFs can improve the price discovery of the underlying assets (Ernst, 2020; Hasbrouck, 2003), the short‐term information efficiency of

underlying assets (Glosten et al., 2021), and the liquidity of underlying assets (Holden & Nam, 2017;Sağlam et al., 2020), and expand the market's ability to

hedge (Huang et al., 2021). In contrast, others have highlighted risks and concerns of ETFs, including more rapid transmission of liquidity shocks, higher

return correlations among stocks held by the same ETFs (Da & Shive, 2018), higher return volatility (Ben‐David et al., 2018; Broman & Shum, 2018), higher

comovement in liquidity of the underlying assets (Agarwal et al., 2019), and elevated market instability (Bhattacharya & O'Hara, 2018).

5

Costs in the European ETF market on arbitrageurs include creation and redemption fees, depending on the rules in the exchanges on which ETFs are

listed (Kaminska, 2011), operational costs in each jurisdiction, and increased risk of settlement failure that arises because the broker's back‐office system

cannot transfer ETFs across central securities depositories (CSDs) in time (Lijnse& Scholtes, 2010).

6

The sample makes a big difference in the direction of relative liquidity. Even in the US ETF market, relative liquidity can be negative when including a

broader sample of ETFs (Box et al., 2021). European ETFs are generally also much more illiquid and less price efficient compared to their US counterparts

(Broman, 2020).

1104

|

JOURNAL OF FINANCIAL RESEARCH

We first explore how liquidity and relative liquidity affect ETF price efficiency in the Chinese market. The

liquidity of ETFs is essential in keeping prices aligned with fundamental value and minimizing price deviations from

the NAV. Several papers have studied the impact of liquidity on price efficiency in the context of US equity ETFs

and demonstrated that lack of ETF liquidity (Cespa & Foucault, 2014; Madhavan & Sobczyk, 2016; Petajisto, 2017)

or a large liquidity mismatch between the ETF and its underlying portfolio (Pan & Zeng, 2019) discourages arbitrage

and leads to more persistent mispricing. Bae and Kim (2020) highlight the impact of liquidity on price efficiency,

showing that ETFs have large tracking errors

7

when underlying assets are less liquid. However, it is unclear whether

these findings translate to the Chinese ETF market, with its unique trading process and market structure and ETFs

that are less liquid than their underlying portfolios. Thus, the impact of liquidity on the price efficiency of Chinese

equity ETFs may differ from that of developed markets and needs to be empirically analyzed.

To study the relations among ETF liquidity, relativeliquidity, and price efficiency, we use daily absolute deviation

and premium (discount) persistence to measure the size and duration of price inefficiency and four measures of ETF

liquidity: relative quoted spread, relative effective spread, market depth, and Amihud illiquidity.Our findings reveal a

strong negative correlation between ETF liquidity and both the magnitude andpersistence of price deviations after

controlling ETF and day fixedeffects in regressions. Consistentwith the literature(Pan & Zeng, 2019), we also findthat

a greater liquidity mismatch between the ETF and its underlying portfolio leads to higher and more persistent

mispricing. Broman andShum (2018) demonstratethat for US equity ETFs,relative liquidity, as opposedto the liquidity

of the ETF itself, is significantly more importantfor short‐term versus long‐term investors. However, Chinese equity

ETFs have not shown sufficient liquidity advantages to attract short‐term traders. In the Chinese market, we find that

the impact of ETF liquidity on price efficiency remains statistically significant after controlling for liquidity mismatch.

ETF prices and NAVs must be up‐to‐date for the deviation to be interpreted as an accurate gauge of price dislocation.

Otherwise, the price deviation is artificial and biased when the ETF closing prices, NAVs, or both are stale. This issue is

well recognized (Broman, 2020;Petajisto,2017).We use ETF closing midquote prices instead of closing prices to address

theconcernofstaleETFclosingprices.ToresolvetheissueofstaleNAVs,weadoptthepeer‐group approach proposed

in Petajisto (2017). We compute the percentage deviation of the ETF closing midquote prices from adjusted NAVs, which

areestimatedfromthepricesofETFsthattracktheidentical benchmark indices. Our results remain consistent after

purging the effect of stale prices from our deviation measures. We also use transaction‐cost‐adjusted deviations to

confirm that the documented inefficiency cannot be explained away after considering arbitrage costs.

Although our ordinary least squares (OLS) regressions control for observable ETF characteristics and include

ETF and day fixed effects, ETF liquidity may be endogenous. A higher ETF price efficiency may attract more active

trades, thereby resulting in better liquidity. To address this concern, we rely on the quasi‐natural experiments

provided by changes in the relevant rules of the Shenzhen Stock Exchange (SZSE) and Shanghai Stock Exchange

(SSE), specifically, the addition of liquidity service providers.

8

According to the new regulations, liquidity service

providers provide market‐making services for some ETFs, which improve ETF liquidity. Our empirical results reveal

that the addition of liquidity service providers does indeed improve price efficiency of ETFs, consistent with a

causal interpretation between liquidity and price efficiency.

Next, we analyze how arbitrage restrictions influence price efficiency. Arbitrage between primary and

secondary markets keeps ETF prices aligned with their NAV. However, there are significant differences between

China and the United States in market regulations and rules. The T+1 trading rule in China, different creation/

redemption rules, and inactive short selling on stocks and ETFs in China all pose direct and serious restrictions on

7

Bae and Kim (2020) define tracking errors as the return differences between ETF price and NAV, NAV and index, and ETF price and index. Deviations

between NAV and index and between ETF price and index are mainly due to fees and expenses or representative sampling of the ETF product structure

and thus may be less reflective of actual price inefficiency.

8

The SZSE and SSE introduced liquidity service providers to provide market‐making services for Harvest CSI 300 ETF and Huatai Berry CSI 300 ETF in

May 2012. Since then, an increasing number of fund management companies have added securities firms as liquidity services providers for their ETFs. The

exchanges rate their market‐making services on a quarterly basis and provide incentives for outstanding market makers, including reductions in

commission fees and refunds of transaction fees. See Section 2.1 for more details.

EFFECT OF LIQUIDITY AND ARBITRAGE

|

1105