Short-term and Long-term Leverage Effect in Volatility Forecasting: Modeling and Analysis Based on GARCHMIDAS Model
| Main Author: | Cheng, Zhenxing |
|---|---|
| Format: | info publication-thesis eJournal |
| Bahasa: | cmn |
| Terbitan: |
, 2019
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/3517856 |
Daftar Isi:
- This is my undergraduate thesis, which mainly refers to Zhi Y. P. and L. Liu. (2018). although it has successfully passed the graduation defense, there are still many problems in this thesis, so it is not recommended as a reference for academic purposes. Abstract: According to the Heterogeneous Market Hypothesis, a financial market is composed by participants having different trading frequencies. The trading behaviors of investors with various time horizons cause different types of volatility components. Therefore, the various types of investors are concerned about whether the different short-term or long-term component of stock return volatility is more strongly related to past negative return, also named as leverage effect. Therefore, it is meaningful to consider the short-term and long-term leverage effects in volatility prediction. This paper extended the GARCH-MIDAS model proposed by Engle et al. (2013). The short-term and long-term leverage effects in the Shanghai 180 Index and Shenzhen 100 Index series are detected by adding corresponding asymmetric terms in the model. The effect of short-term leverage effect is that negative returns in the past can cause greater short-term fluctuations than positive returns in the future. The effect of long-term leverage effect is that positive returns in the past will increase long-term fluctuations in the future, while negative returns will reduce long-term fluctuations in the future. This paper also finds that the addition of leverage effect can improve the accuracy of the prediction of volatility, which is mainly attributed to the short-term leverage effect. In the estimation method, this paper adopts the rolling window technology. By fixing the monthly window and moving by day, the model can contain more information about long-term fluctuations. This improves the persuasiveness of the model. In the evaluation method, six popular loss functions and success indicators are used to avoid the evaluation errors caused by a single standard. Keywords: volatility prediction; short-term leverage effects; long-term leverage effects; GARCH-MIDAS
- As my first academic attempt, there are many problems in this paper. I just want to make a memorial here.