Climate Disaster Risks—Empirics and a Multi-Phase Dynamic Model†
Round 1
Reviewer 1 Report
The paper is interesting but a bit difficult to read since you develop lot of ideas and two different parts mixing both empirics and IAM model. You should simplify, reduce the size of the paper and just keep the stronger results.
Table 1 : only 41 observations, in what extent it is relevant ?
Page 15 : why not use Engle and Granger sine you have only 2 variables and one cointegration relationship at maximum ? You should use cointegration tests with small samples corrections.
What is the value added of the empirics ? In the way with Thomas and Lopez, 2015, so what ?
If limited number of observations is a problem for Johansen, it is also the case for ADF and KPSS tests. Why did not use DF-GLS test that is more powerful than ADF and KPSS tests? What about some potential breaks leading some potential bias? A minima Zivot Andrews (1992) test can be used. In appendix, you present only results with deterministic trends, why? Are you sure deterministic trends are significant?
What is the exact definition of a climate disaster? Authors use climate disasters but it is difficult to cleacly understand the scope of this term. It is however crucial to understand the demonstration of the paper. Coronese et al. definition? Is it possible to add droughts to wildfire storms or extreme temperatures for example? Are the transmission channels to the economy the same? Can you detail them here? We note that the dynamics within these categories (it is interesting besides retailers in the paper) are different: wildfires are few (but they can have very strong effects as recently in Australia), so the dynamics are very random and it is difficult to identify a trend, in what extent is it relevant to study them?
page 11: “Are very limited but central Europe and Asia”: sentence seems to be not complete…
CO2 data are collected from Lao Laboratory, why? What are the differences with the well-known CAIT database for example?
Though the “empirics” part is interesting, especially the figures about climate disasters dynamics, I am not convinced by the robustness of the regressions. If there is no cointegration, the estimated equation (2) can lead to spurious regression results? I have also some doubts about equation (3), all parts of the equations should be stationary, what about CO2 growth? Indeed, there is a debate about CO2 dynamics: TS or DS?
Why did you use a Poisson estimator? Lot of zeros observations? You should motivate the econometric assumptions in this part of the manuscript.
What correction did you use for “Fixed-effects Panel with heteroskedasticity and serial correlation robust standard errors”, clustering?
I have some difficulties to understand the bridge between the empirics part and the model part, please give some motivations and details in your manuscript. It looks like two orthogonal parts…
In a more global way, the paper sets lot of very interesting questions, problematics, ideas but with a very important density of ideas, sometimes not well ordered. Maybe should simplify a bit the paper and/or reduce the number of ideas you want to outline to improve their understanding by the reader. In addition it is not easy to assess the net value added of the paper.
What is the difference between your model and Maurer et al. (2018)? You model is in the spirit of IAM with 5 state variables but what is the disaster shock here? The nature of the shock? If there is no shock (I guess), how you link the empirical part and the model?
Some references related the literature about persistent effects of climate on productivity are missing : see for example Letta and Toll (2017), ERE. Sandra Batten (2018), Climate and The Macroeconomy, Bank of England.
Concerning the model part, see Gallic and Vermandel, European Economic Review, Weather Shocks, forthcoming
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Thank you for the opportunity to review the paper entitled "Climate Disaster Risks – Empirics and a Multi-Phase Dynamic Model". This manuscript covers a very important issue on climate disaster and macroeconomic policy needs. I feel that this manuscript provides the necessary background information and refers to all the related literature. More importantly, theoretical model of disaster shock is developed and research method is clearly described. In particular, results of Poisson regressions as well as linear fixed-effects panel regressions provide a very interesting insight to readers. They presented results of the three-phase model and showed the dynamic paths. I fully support publication of this manuscript. My only suggestion to the authors is to improve the writing of some complicated and long sentences. Also, minor spell check is required. Some of the literature cited in the text might be missing in the references.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments on "Climate Disaster Risks -- Empirics and a
Multi-Phase Dynamic Model"
The design of an appropriate and feasible climate policy to support mitigation and adaptation actions is an important research question. The main contribution of this paper is to suggest a three phase model with endogenously determined transition from fossil fuels to renewable resources for energy production.
My comments and suggestions are:
- The paper starts with a lengthy exploratory data analysis (pages 5 to
18, and appendices A.1 to A.3), presumably to convince the readers that there is a long-run connection between CO2 emissions and climate related disasters. I do not think that readers of this journal will have any doubt. I do not think plotting counts of different types of disasters for different regions of the world in Figures 1 to 6 and related discussions in the text add anything to the paper. Also, as the authors correctly state, there is no variation in CO2 concentration across regions, so the point of using a regional panel, given that the theoretical model of the paper is not a regional model, is lost to me. I think a small section that explored the dynamic correlation between total disasters in the world and CO2 levels would be sufficient as a motivation in this paper. That would need two or three pages maximum, and will leave more room for a clearer exposition
of the three phase model. - The "most recent 17 increases in CO2 concentration" was ambiguous to me. Do you mean a one sided moving average of order 17 in changes in CO2 or a one sided moving average of variable length that only uses positive changes and goes back as far as needed to find 17 positive changes (i.e. 17 increases)? I guess that this is a smoothing filter, but it sounds strange and arbitrary.
- The theoretical model needs more details. Appendix A.4 can perhaps be incorporated into the main text to make the model more transparent.
- The choice of parameters must be explained. For example, footnote 14 states that "given the quality and heterogeneity of the data it is very hard to undertake more precise parameter calibration." However, there is no mention of what data were used in calibrating the parameters. If these parameters are not estimated, then the authors must discuss the sensitivity of the results to choices of parameters.
On the whole, while I am very happy that the authors are thinking about these important issues, I believe that the main part of the paper (the multi-phase dynamic model) was not explained in a way that an interested reader could reproduce its results.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I read the response to the referee's letter as well as the new version of the paper and I now recommend to accept the paper for publication considering the authors have answered to all comments and make the relevant corrections.
Reviewer 3 Report
The authors have addressed my comments adequately in their revision. I have no further comments.
