Machine Learning Applied to Banking Supervision a Literature Review

Round 1

Reviewer 1 Report

The paper is easy to read but falls short on several fronts: First, while the authors argue that this is only a lit review, a pure literature review paper should be more comprehensive than this.  Second, the authors pick and choose papers for the literature review, which is fine in a paper like this, but it is too early for central bank supervision point of view for a paper like this.  I really see where the authors are trying to go with this paper, but I feel they fall short of their goal. 

Author Response

We thank the reviewer for his/her review and the valuable comments.

In the revised version of the paper, we took into account the reviewer’s comments. In particular, we included a section concerning related works, a section discussing the dataset used in this area (machine learning in the context of central banks), and an in-depth discussion covering more papers.

Finally, we strengthened our view on the importance of this topic: while the reviewer thinks this contribution came at an early stage, we explained our reasons that are based on the experience of the authors with central banks. In particular, the first author works for the central Bank of Portugal and the European Central Bank. As a professional in this field, he sees a clear contribution for fostering further research in this area and for supporting the research that is currently been developed within the central banks. Thus, we believe this paper is timely and could serve as a solid reference for all the researchers and/or practitioners that want to apply machine learning methods in the context of banking supervision.

Reviewer 2 Report

The paper is interesting and has the potential to become a very good contribution. However, I suggest the authors to extende the spectrum of examined papers to make their work an effective review of the ML contribution to address the issue of banking supervision.

Author Response

We thank the reviewer for his/her review and the valuable comments.

In the revised version of the paper, we took into account the reviewer’s comments. In particular, we included a section concerning related works, a section discussing the dataset used in this area (machine learning in the context of central banks), and an in-depth discussion covering more papers. Moreover, we added more Figures and Tables to summarize the information collected in this study.

Finally, we strengthened our view on the importance of this topic: we believe this paper is timely and could serve as a solid reference for all the researchers and/or practitioners that want to apply machine learning methods in the context of banking supervision.

Reviewer 3 Report

The authors provide a literature review about the machine learning applications for banking supervision. The work is interesting and the methodology is explained well. However, the paper does not include a "Related Work" section describing similar works. Thus, the contributions of the paper with respect to other works are not clear. Although the paper constitutes a literature review, there can be similar works. Moreover, the various papers studied can be analysed further with more details. Also, they can be organised better with more fields with respect to the tables. These fields can relate technical details, such as what type of ML models are used. Moreover, the purpose of each paper can be used. Finally, relevant datasets should be also noted. Finally, the paper should be checked entirely about possible writing errors and typos.

Author Response

We thank the reviewer for his/her review and the valuable comments.

In the revised version of the paper, we took into account the reviewer’s comments. In particular, we included a section concerning related works, a section discussing the dataset used in this area (machine learning in the context of central banks), and an in-depth discussion covering more papers. Moreover, as suggested by the reviewer, we added more information in the Tables in the Appendix to summarize the information collected in this study.

We also strengthened our view on the importance of this topic: we believe this paper is timely and could serve as a solid reference for all the researchers and/or practitioners that want to apply machine learning methods in the context of banking supervision. Finally, we checked the paper for possible writing errors and typos.

Round 2

Reviewer 2 Report

The paper offers a comprehensive analysis cocerning the way machine learning (ML) techniques have been applied to the issue of risk assessment in banking. The authors searched along main engines including Google Scholar, Springer Link, and ScienceDirect and focused on the articles including the search terms “machine  learning” and (“bank” or “banking” or “supervision”). The analysis and discussion is now improved. Please note that Fig. 4 is missing. Overall I think that the improvements make the article ready for publication.

Author Response

We thank the reviewer for his/her comments. In the revised version of the paper, we fixed this issue by including Figure 4.

Reviewer 3 Report

The authors covered most of the comments raised during the previous review process. However, it still presents some significant flaws. First, Fig. 4 does not exist! Secondly, as mentioned during the previous review process, each work can be analysed further, discussing also the evaluation details. The authors could also discuss further the various datasets. Finally, the paper should be re-checked about potential writing errors and typos.

Author Response

The authors covered most of the comments raised during the previous review process.

R1. We thank the reviewer for his/her comments.

Fig. 4 does not exist!

R2. In the revised version of the paper, we included Figure 4. 

Each work can be analysed further, discussing also the evaluation details. The authors could also discuss further the various datasets. Finally, the paper should be re-checked about potential writing errors and typos.

R3. In the revised version of the paper, the Tables in Appendix have been extended to include more details on each paper, results, and the respective datasets. We also checked the paper for typos.

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

An excellent paper.  Just a few suggestions:

105 ‘they would merit from’ sounds wrong to me

269 ‘This type of models’ -> ‘These types of model’ or ‘This type of model’

281 ‘Random Forest’ -> ‘a random forest’

283 ‘in their future path’ -> ‘on their future path’

419 ‘one of the most recently works’ -> ‘one of the most recent works’

420 ‘day to day’ -> ‘day-to-day’

Reviewer 2 Report

Comments in file below

Comments for author File: Comments.docx

Reviewer 3 Report

Ignore my other evaluation, I just had to choose something.

This paper does not need to be evaluated for this journal, because it does not fit the scope of this journal at all:

"Applied Sciences is an international, peer-reviewed, open access journal on all aspects of applied natural sciences"

Banking does not belong to natural sciences.