Does investor sentiment dynamically impact stock returns from different investor horizons? Evidence from the US stock market using a multi-scale method

ABSTRACT

This article uses the investor sentiment index to investigate the Granger causality between investor sentiment and stock returns for the US economy using a multi-scale method. To focus on the local analysis of different investor horizons, bivariate empirical mode decomposition is used to decompose time series of investor sentiment and stock returns at different timescales. We employ the linear and nonlinear integrated Granger causality method to examine the causal relationship of decomposed series on similar timescales. The results indicate both strong bilateral linear and nonlinear causality between longer-term investor sentiment and stock returns. However, there is no strong evidence for correlation of stock returns and investor sentiment on shorter timescales.

KEYWORDS:

• Bivariate empirical mode decomposition
• investor sentiment
• stock returns
• Granger causality

JEL CLASSIFICATION:

• C14
• C22
• C32
• G10
• G11

Acknowledgements

This work is supported by grant from Guangdong discipline construction project (No. GD16XYJ12).All authors are grateful to the editor and the anonymous reviewer for suggestions and comments. All errors are ours.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Due to limits of space, this article does not include the model derivations defined by Rilling et al. (2007) and Kyrtsou and Labys (2006). Please refer to the original papers for further details.

² Available at: http://www.wind.com.cn/.

³ Available at: http://people.stern.nyu.edu/jwurgler/.

⁴ See Yu et al. (2015) for the specific explanation of the different decomposing timescales.

⁵ The ADF test results are shown in Table 1. We adapt the GLS-DF test and the Breitung unit root test to obtain similar results, and all show the all series are stationary (due to space limitations, these data are not presented).

⁶ Empirical evidence has indicated that the GJR-GARCH (1,1) model proposed by Glosten, Jagannathan, and Runkle (1993) can be used to conduct a good estimation for the fluctuation of financial variables (Jiang, Nie, and Monginsidi 2017).

Additional information

Funding

This work is supported by grant from Guangdong discipline construction project [No. GD16XYJ12].