Stress Testing and Systemic Risk Measures Using Elliptical Conditional Multivariate Probabilities
Round 1
Reviewer 1 Report
The paper can provide great motivation and theoretical background for many researchers to study the CoVaR on the elliptical distribution. I would ask some minor questions about the topic. Can the idea extend to the Generalized Hyperbolic distribution class which is an extension of the student-t distribution? Also, I recommend to refer the following paper which discussed the practical application of CoVaR to portfolio optimization:
T. Kurosaki, Y. S. Kim (2013), Systematic Risk Measurement in the Global Banking Stock Market with Time Series Analysis and CoVaR, Investment Management and Financial Innovations, 10(1), 184-196
Author Response
I thank this reviewer for the insightful and useful comment that helped to improve the manuscript. I added in the conclusion a final comment on the Generalized Hyperbolic distribution class which is indeed a good suggestion for future work. I also added the reference.
Reviewer 2 Report
Although most derivations may be well known to readers with a statistical background the paper is coherent and I like that.
I would simply suggest, perhaps, to motivate a little bit better the empirical application.
Well done!
Author Response
I thank this reviewer for the insightful and useful comment that helped to improve the manuscript. I added some wording of motivation on the empirical application stressing that it is an example.
Reviewer 3 Report
1) The notion of "stress" and "systematic risk" should be defined.
2) The novelty or the new value added of the research should be emphasised.
3) "Mahlanobis impact factor" should be defined/described shortly.
4) "In this paper, I investigate the vast family of multivariate elliptical distributions" Unfortunately, I cannot find any Table summarizing the outcomes or conclusions for this "vast family". The aforementioned statement of the author seems like we should find in the paper numerous examples, but only 2 distributions are examined in details in the text.
5) There is no information which software was used in the paper.
6) I think the paper would be greately improved and its usability for other researches would be much higher, if the source code for computations and data would be appended to the paper.
Author Response
I tank this reviewer for his/her valuable comments. I have modified the manuscript accordingly with the suggestion. Specifically points are replier inline.
1) The notion of "stress" and "systematic risk" should be defined.
I added a paragraph at the beginning of ether introduction defining these consents.
2) The novelty or the new value added of the research should be emphasised.
I expanded a bit the part where I compare with the literature mainly pointing out that there are no available measures of multivariate risk propagation.
3) "Mahlanobis impact factor" should be defined/described shortly.
I have better defined/described and cross-referred it now.
4) "In this paper, I investigate the vast family of multivariate elliptical distributions" Unfortunately, I cannot find any Table summarizing the outcomes or conclusions for this "vast family". The aforementioned statement of the author seems like we should find in the paper numerous examples, but only 2 distributions are examined in details in the text.
This is not correct. The three main measures introduced in the bodyy of the paper are general and apply to the whole family. Th three in the appendix instead are specifically described for the normal and Student-t case. I know better clarify this point.
5) There is no information which software was used in the paper.
I use a Matlab code. I have now put it in a ghithub repository: https://github.com/financialcomputing/Systemic-Risk-Measures
6) I think the paper would be greately improved and its usability for other researches would be much higher, if the source code for computations and data would be appended to the paper.
See the ghithub repository: https://github.com/financialcomputing/Systemic-Risk-Measures
