Predicting Volatility Based on Interval Regression Models
Round 1
Reviewer 1 Report (Previous Reviewer 2)
The results based on RV (realized volatility) should be included, and details about RV should be provided (data source, RV measure construction, etc.).Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report (New Reviewer)
The interval regression method presented in this paper is an interesting proposal for modeling time series with high volatility. The literature review and the description of the method are sufficient. The empirical research focuses on the S&P500 index and covers periods of several crises: European debt crisis, China’s abnormal market fluctuations, the trade disputes between China and the United States, and the COVID-19 pandemic. Unfortunately, the empirical research does not include the period of the global financial crisis, which had a significant impact on the US market.
Author Response
The interval regression method presented in this paper is an interesting proposal for modeling time series with high volatility. The literature review and the description of the method are sufficient. The empirical research focuses on the S&P500 index and covers periods of several crises: European debt crisis, China’s abnormal market fluctuations, the trade disputes between China and the United States, and the COVID-19 pandemic. Unfortunately, the empirical research does not include the period of the global financial crisis, which had a significant impact on the US market.
Response:
Thanks for this comment.
Actually, our empirical data spans from January 3, 2006 to December 30, 2020, including the period of the global financial crisis. Thus, our in-sample fit and prediction results (Tables 2-5) are based on the data including the global financial crisis period. As mentioned in the discussions for Figures 1-4 (Lines 342-346), "It can be seen that the smoothed regime probabilities of regime 1 are large around 2009, 2012 and 2020, which correspond to the financial crisis in 2008, the European debt crisis in 2011 and the stock market shocks caused by COVID-19 in 2020. This shows that the fitted Markov regime switching structure measures the fluctuations of the S&P 500 index accurately."
Reviewer 3 Report (New Reviewer)
I find this paper to be very interesting. It is generally well-written and I believe it makes a contribution to the financial econometrics literature on volatility modeling.
I have a few comments that I would like you to address.
1. (line 187) "Thus the CRM-H model can separate the contributions of investors trading at different frequencies to future volatility."
This statement doesn't make sense to me. For instance, suppose trader A trades weekly. Then trader A's activity potentially contributes to Rrt-1,t-5 and to Rrt-1,t-21. Also, how does trading in the past contribute to future volatility? The AR structure of volatility does not necessarily imply a causal relationship in which past vol contributes to future vol. Another way of looking at it is that exogenous factors (eg. uncertainty about future inflation) cause volatility in equity returns and the effects tend to persist for a while resulting in an AR structure in volatility. I would suggest that you just delete this sentence.
2. (line 211) You should specify 0<=P00,P11<=1.
3. (footnote 6) I would suggest deleting "with 236 citations in Google Scholar" since it's not necessary to mention and this number will likely increase over time anyway.
4. (line 627) "...which assists investors’ more profitable asset allocation and derivative trading activities."
You haven't actually shown how your model improves the outcome of any specific trading activity, so it's not really a justified statement. It's probably a fair assumption that anyone reading this paper understands the importance of volatility modeling in financial markets.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report (Previous Reviewer 2)
The authors have satisfactorily addressed my concerns.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
The paper has some minor issues in my opinion, and after they are resolved, I think that the paper can be published.
Introduction should have a better transition from the first paragraph in which you describe the intro to volatility and the second paragraph in which you describe what you do in your approach. There is a missing link in which you set the scene of what are research gaps, that you are filling. You have this in the intro, but after your second paragraph. Introduction section should be re-ordered so that the flow is a bit better.
Another thing is that the intro is too long. It should be divided into a true intro (question, gap, contribution, etc.) and the rest should be a literature review section. As of now, due to it being so long, you almost forget the point as you describe many other research. If someone is looking at your intro, it should almost have all needed info about what the paper is about without looking at the detailed analysis of the literature review section.
So please rewrite the intro into 2 sections. Otherwise, the info you put there is OK.
Methodology section: formulas need to be a bit tigther, as now they are a bit messy, some notation description is missing, e.g. epsilon in formula (4) , etc. Although we know what some of them are, they should be defined as the rest of the symbols, so please check this throughout the paper.
Some tables fall out of the page (e.g. table 4).
Now, the problem in the interpretations section, i.e. the empirical one, is that you do not compare your results to related literature, what are the similarities and differences in the results and why. This is important for those who consider all these approaches of volatility estimation and prediction, to see what can be done and used in practice. Although you compare the variants of models you described in the methodology section, there aren't comparisons that I have mentioned.
Another section should be put before the conclusion, i.e. it can be a sub-section named Discussion, in which you summarize the findings, not in a statistical manner of what measure is better or worse, but rather summarize all of the findings in terms for potential investors. What is best and why, what is worst and why. How to use this in practice, in portfolio management. Due to transaction costs, what could be some downfalls? Some things from conclusion regarding these summaries should be put here.
Then, the conclusion should be what now, after your results and recommendations for investors. Please put here the advantages of your approach, and shortfalls of the study, with recommendations for future work, what should be done next.
Reviewer 2 Report
See attached file.
Comments for author File: Comments.pdf