Do Bubbles Alter Contributions to Price Discovery? Evidence from the Chinese Soybean Futures and Spot Markets

ABSTRACT

The purpose of this study is to investigate whether bubbles have significantly influenced the performance of price discovery in Chinese soybean futures markets. To evaluate the performance of price discovery, we employ forecast error variance decomposition based on structural vector autoregression and directed acyclic graphs, using daily data from six price series of futures contracts expiring at different months and one spot soybean price for the past 13 years. Then, the full-period data are divided into bubble and non-bubble periods via the supremum augmented Dickey–Fuller test. The methods utilized on the full sample for price discovery are conducted again on two subperiods. Our empirical results show that price discovery in the Chinese soybean futures market performs much better in bubble periods and worse in non-bubble period compared to the full period, indicating that bubbles have significantly influenced the performance of price discovery in Chinese soybean futures markets.

KEY WORDS:

• Bubbles
• directed acyclic graphs
• price discovery
• structural vector autoregression

Notes

1. Because of space limitations, we present only a brief introduction to DAG; for a more detailed discussion, see Spirtes, Glymour, and Scheines (2000) and Bessler and Yang (2003).

2. Here the Chinese soybean futures refer to the No. 1 soybean futures. No. 1 soybean contracts are non-genetically modified soybeans.

3. Heilongjiang is the main production area for soybean and other regions is heavily dependent on imported soybeans.

Additional information

Funding

This work was supported by the Natural Science Foundation of China [71771101, 71501079, 71673103]; the Fundamental Research Funds for the Central Universities [2662017JC050].