Institutional high frequency trading and price discovery: Evidence from an emerging commodity futures market

• Author: Difang Wan,Yue Zhao
• Published date: 01 February 2018
• Date: 01 February 2018
• DOI: http://doi.org/10.1002/fut.21888

Received: 19 October 2016

|

Accepted: 24 September 2017

DOI: 10.1002/fut.21888

RESEARCH ARTICLE

Institutional high frequency trading and price discovery:

Evidence from an emerging commodity futures market

Yue Zhao

|

Difang Wan

School of Management, Xi’an Jiaotong

University, Xi’an, Shaanxi, China

Correspondence

Yue Zhao, PhD, School of Management,

Xi’an Jiaotong University, 28 West

Xianning Street, Xi’an, 710049 Shaanxi,

China.

Email: zhaoyue_xjtu@foxmail.com

Funding information

National Natural Science Foundation of

China, Grant numbers: 71173166,

71373202, 71671138

We compare the effects of institutional and individual trading on intraday price

processes in the emerging c ommodity futures market of Chin a with a unique

trade-by-trade dataset. Institutional investors collectively facilit ate price discovery

with positive permanent price impacts,but their beneficial role is time agglomerated,

that is, only institutionalhighly-concentrated trades executed at the same millisecond

are accompanied by information effects. Transitory price disturbances are mitigated

by informed institutional highly-concentrated trading in the agricultural sector,

whereas these disturbancesare alleviated by liquidity-enhancing individualtrading in

the industrial sector. Overall,the entire market is abnormally dominated by transitory

volatility instead of informational volatility.

KEYWORDS

investor structure, price impacts, trading intensity

JEL CLASSIFICATION

G13, G14

1

|

INTRODUCTION

Using a unique trade-level dataset from the Dalian Commodity Exchange (DCE) in China, this study investigates the

relationships between intraday price processes and trading activities of two distinct groups: institutional investors and individual

investors. Furthermore, we compare ultra-high-speed trading at the same millisecond against all other trading within each group.

Unlike developed markets, the emerging commodity futures market in China is highly dominated by individual investors,

who account for approximately 90% of the entire market. This unbalanced investor structure has caused excessive short-term

speculative trading by both institutional and individual participants. Given that trades involving institutions make up only a tiny

proportion of the market and are characterized by strong speculative intentions, it is compelling to examine the role of

institutional trading on market efficiency. The dataset identifies institutional proprietary accounts and individual accounts to

which the buying and selling activities belong. Therefore, we are able to examine the effects of institutional trading (INST) and

individual trading (INDV) on price processes separately and directly.

Moreover, the launch of electronic trading platforms and the extended use of automatic trading terminals have facilitated

market liquidity by increasing trading capacity substantially while reducing execution time dramatically. However, it has also

aroused widespread concern that the explosive growth of high-speed and profit-driven trading, such as high-frequency trading

(HFT), might distort the informativeness of the price and impede the stability of the market, which in turn would leave more

space for speculation or even manipulation, particularly in emerging markets. In the detailed dataset, the elapse of time between

any two trades is measurable to the minimum of a millisecond. When trades are executed by the same type of investors,

institutional, or individual, at the same millisecond, they are supposed to possess stronger continuity on the time dimension and

J Futures Markets. 2018;38:243–270. wileyonlinelibrary.com/journal/fut © 2017 Wiley Periodicals, Inc.

|

243

signify a higher level of trading intensity. Therefore, whenever adjacent institutional trades have a zero-millisecond time

interval, they are defined as institutional highly-concentrated trading (INST

C

), which signifies ultra-high trading frequency

within an indivisible unit of time. To facilitate comparison, the rest of institutional trades are defined as institutional ordinary

trading (INST

O

) since they can be independently identified on the time dimension. Moreover, individual investors are also able to

trade intensively within an indivisible millisecond due to their predominance in the emerging market. Accordingly, individual

highly-concentrated trading (INDV

C

) and individual ordinary trading (INDV

O

) are defined in the same way within the category

of individual trading. Hence, we are able to subdivide institutional and individual trading according to the time interval between

trades and examine the impact of each category on price processes as dependent on the trading intensity.

The pre-trade return, which is calculated as the difference between the price of a given trade and the price of a prior trade and

then scaled by signed trading volume, is used as a proxy to detect the role of liquidity provision versus consumption. Negative

pre-trade return is associated with the willingness to long (short) while the market is falling (rising), thereby indicating the role of

liquidity provision. Conversely, positive pre-trade return is consistent with the desire to buy (sell) during market ups (downs),

therefore implying the role of liquidity consumption. However, it is insufficient to depend solely on the pre-trade return, because

the liquidity-enhancing effect of trades where both the buy side and the sell side are from the same category (sub-category) has

been balanced out and thus neglected. The market share of unsigned trading volume as a traditional measurement of contribution

to liquidity thereof should be combined. Based on both pre-trade returns and market shares, we conclude that liquidity is mostly

consumed by institutional highly-concentrated trading, and it is individual trading that mainly benefits the market with surplus

liquidity.

Following the research of Brogaard, Hendershott, and Riordan (2014), we use the state space model (SSM) to decompose

observed price series into permanent and transitory price series. The permanent components represent the unobserved efficient

prices. Innovations in efficient prices are defined as permanent price changes and are supposed to be caused by new information

on the fundamental value of the underlying commodity. The transitory components represent unobserved price errors, which will

be corrected later.

Both permanent price changes and transitory price errors are also related to trading activities, which are measured using

signed trading volume.

1

Trading along with permanent price changes can be interpreted as incorporating information into prices,

thereby enhancing market efficiency. Conversely, trading against permanent price changes is information-biased and vulnerable

to adverse selection. Trading along with transitory components introduces temporary distortions into prices and consequently

increases short-term volatility. In contrast, trading against transitory components mitigates temporary price disturbances.

Institutional trading collectively plays a beneficial role in price discovery by trading along with permanent price changes.

Given market clearing, overall individual trading is in the opposite direction of efficient price movements. These findings

confirm that institutional investors are more sophisticated and better informed, even when they account for only a tiny portion of

the emerging commodity futures market. In contrast, the vast majority of market participants, individual investors, who provide

liquidity to the market, are unable to predict the movements of efficient prices. As a result, their orders are at the risk of being

adversely selected.

When institutional trading is further divided into highly-concentrated trading and ordinary trading, only institutional highly-

concentrated trading is consistent with permanent price changes, whereas institutional ordinary trading is contrary to permanent

price movements. Thus, the information advantages of institutional investors have a marked tendency for time agglomeration,

which casts doubt on their ability to gain and maintain the benefits on information.

The disaggregate model for individual trading reveals significant negative relationships between permanent price changes

and trading activities, regardless of trading time intervals, which confirms the uninformed role of individual investors.

Although not negligible, the contribution of institutional investors to price discovery is very limited. The positive role of

institutional trading is severely restricted to the small size of their market share. Information contained in informed institutional

trading is easily diluted or even overwhelmed by the huge volume of uninformed trading. Moreover, institutional investors

cannot behave independently without interacting with other market participants. Surrendered to the dominance of uninformed

trading, their judgments on price would be disturbed or even misled, either irrationally or under the pressure of some liquidity

needs. This limitation is revealed by associating trading activities with transitory price errors.

Overall, institutional trading also disturbs prices by trading in the same direction of transitory price errors. After separating

trading activities based on trading intensities, it is noticeable that institutional ordinary trading always impairs price efficiency

with transitory price impacts. Given simultaneously negative permanent price impacts, institutional ordinary trading is inclined

to be noise-oriented.

1

For consistency, innovations in signed trading volume are used when modeling permanent price changes.

244

|

ZHAO AND WAN