How oil price shocks affect investor sentiment: new evidence from China

ABSTRACT

This paper studies the effect of structural oil price shocks on China’s investor sentiment using the Bayesian inference structural vector autoregression (VAR) model by Baumeister and Hamilton (2019). We find that oil supply shocks followed by consumption demand shocks have positive and persistent effects on China’s investor sentiment, whereas aggregate and inventory demand shocks have only temporary effects. The effects of these shocks are transmitted mainly by affecting investor’s confidence in domestic economic fundamentals and the international economic environment. Furthermore, oil price shocks also induce different trading motivations in the stock market.

KEYWORDS:

• Oil price shock
• investor sentiment
• bayesian inference
• structural VAR

JEL CLASSIFICATION:

• C23
• Q42
• Q51

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ We use Kilian’s index of global economic activity accompanied by his three-oil-shock-source model as a robustness check.

² Note that we also use the investor sentiment proxy constructed through the principal component analysis method from Cheema, Man, and Szulczyk (2020) as a robustness check.

³ For the detailed discussions about the BHSVAR model, please refer to Section 2 of Baumeister and Hamilton (2019).

⁴ We do not find a significant impact of supply shock on economic activity shown in Baumeister and Hamilton (2019). The reason might be that our sample period is not so long for the accumulation of the significant role of oil supply.

⁵ In theory, it is evident that the price of oil would increase when oil supply declines. However, whether oil supply shocks can explain much of oil price fluctuations in history is an empirical question. The contrasting findings in Kilian (2009) and Baumeister and Hamilton (2019) are due to different econometric methods for estimations. The debate about the two approaches is still ongoing. See Kilian and Zhou (2020) and Baumeister and Hamilton (2020). Note that we also use Kilian’s VAR model for robustness check of our estimations.

⁶ The symmetrical interpretation is valid because whether we normalize the shock to be positive or negative is immaterial. We thank Kilian and Hamilton for confirmation of this point.

⁷ Note that the results concerning the eight component indices are also consistent but not reported here for simplicity. They are available from the authors upon request.

⁸ We thank the referee for suggesting this robustness check. The data on the new investor sentiment index (STM) are calculated by Cheema, Man, and Szulczyk (2020) and are available at https://data.mendeley.com/datasets/tzzfcstt9t/2. The sample period of STM is from April 2008 to September 2017.

Additional information

Funding

We thank the funding provided by the Chinese Academy of Social Sciences Youth Startup Project (2020YQNQD007) and the Chinese Academy of Social Sciences Peak Strategy Project “The Advantageous Discipline (Industrial Economics)”.