The inequality labor loss risk from future urban warming and adaptation strategies

Heat-induced labor loss is a major economic cost related to climate change. Here, we use hourly heat stress data modeled with a regional climate model to investigate the heat-induced labor loss in 231 Chinese cities. Results indicate that future urban heat stress is projected to cause an increase in labor losses exceeding 0.20% of the total account gross domestic product (GDP) per year by the 2050s relative to the 2010s. In this process, certain lower-paid sectors could be disproportionately impacted. The implementation of various urban adaptation strategies could offset 10% of the additional economic loss per year and help reduce the inequality-related impact on lower-paid sectors. So future urban warming can not only damage cities as a whole but can also contribute to income inequality. The implication of adaptation strategies should be considered in regard to not only cooling requirements but also environmental justice.

2) Very ambitious adaptation strategies only help to reduce this loss in worker productivity by 5 (or 10%) A possible or even likely reason is that the authors did not reflect about their results this way. If true, this substantially erodes my trust in how rigidly and carefully the analysis has been conducted and interpreted.
There are major results that are unrecognized (or unreflected), which seem to be in conflict with literature and which may point to a flaw in the data analysis and interpretation. Also, the results are taken from just one climate change scenario, without a sensitivity analysis and the adaptation strategy scenarios are highly unrealistic. Moreover, crucial aspects are not covered by the analysis. What is the cost-effectiveness of the adaptation measures? The authors seem to implicitly assume no cost, but that is of course not the case. How will technology and the economy react towards increases in heat-stress? What is the sensitivity of results coming from the model's and scenario's parameters and assumptions? I therefore doubt that this is a rigid analysis, which probably requires more than a major revision. General remarks: Please use a professional English language editing service. The manuscript is difficult to understand in its content, which to a large part is owed to poor language and syntax. The reader is often left guessing what the authors' intended to say. This makes it quite difficult to understand the details of the analysis. There are also many typos, which can be easily avoided using spell check Remaining major comments: 1. The study's main contribution, beside its empirical findings, is on how in detail the economic structure and population residence is accounted for when analyzing the labor induced heat stress. This is indeed a valuable contribution, but quite technical and not well highlighted. It also does not seem to have a major impact on the results (the % bias it corrects is not explicitly reported, it seems to be less than 10%). 2. China may certainly be a very exciting case study, but the reader does not really see why. This could be much better motivated at the beginning (e.g., X% of the world's urban population live in China) 3. The study does not discuss or consider how cost-effective the adaptation strategies could be. This is a major shortcoming, but could be rectified. 4. I appreciate that the authors try to understand how heat-stress impacts different economic activities at the (sub-)urban level instead of the regional or even national scale. Yet, it remains unclear how the authors estimated the location of the economic activities and how exactly they have treated the seasonality of work loads (e.g., construction workers may not work during the midday heat). Also, such a contribution (while definitely valuable) may not be sufficient for a consideration in a top multidisciplinary journal. This is also backed by the rather simplistic consideration of other dimensions (economic adjustments are absent, reliance on one SSP, RCP and exposure function, etc). 5. The economic estimation is very simple and static. There is no endogenous adaptive behavior. Also, no changes in the intensity of the individual economic activities (or even technological improvement). However, the Chines economy is growing very rapdily and technologies are evolving, so is it reasonable to still assume the same economic structure of today? 6. The discussion of how the analysis accounts for changes in the urban population are very difficult to comprehend. I think the authors at least once confuse "urbanization rate" with "share of total population living in urban areas" -see line 236. Minor comments: 1. Defining transportation, manufacturing, etc. as medium work-intensity may overestimate heatstress (also many of these activities can or are done in AC environments -especially until 2050) 2. I think the term "labor support" is not helpful for an interdisciplinary audience. I would also recommend to speak of "losses in labor capacity or productivity" than just "labor loss".
This study assesses the heat-induced impacts on labour productivity and associated economic costs under different adaptation measures at a sub-urban scale. The results from this impact assessment provide some novel and useful insights. Especially, I find interesting the combination of a regional climate model with an urban canopy model to investigate the effectiveness of different adaptation options in urban areas. I have a few questions and improvement suggestions: 1) In Lines 205-206, the authors write: "First, we established only one future climate-society scenario, the modelling results would be slightly different if we compared the results from scenarios under different assumptions". I am not sure that the results would be 'slightly' different. The uncertainties associated with climate modelling (i.e., climate sensitivity) are large, so are socio-economic uncertainties. I guess the authors did not have enough capacity to conduct a dynamic downscaling for multiple RCPs and climate models. However, I wonder if it makes sense to consider, two or three SSPs to investigate the relevance of socio-economic uncertainties.
2) I am a bit surprised that the economic cost from construction is so large compared to other sectors. I understand that construction is more labour-intensive and requires working outdoors, but I guess much more people are working in manufacturing and services. So, I would expect that the total cost of heat-induced impacts on labour productivity in manufacturing and services would be larger than in construction. On that regard, I think it would be useful to have a table or figure showing the sectoral wage rates and the number of people employed in each sector and region. Also, please report the recent penetration of air conditioners by regions.
3) I would expect that the economic structure will change over time (i.e., more income will be generated from the service sector compared to manufacturing and construction). However, it seems that a structural change in the regional economies is not implemented in this study. In this case, this should be mentioned as a limitation. 4) Also, there are other adaptation measures, which could be relevant to urban areas. For example, mechanization in construction, shifting working hours, acclimatisation etc. The authors should mention this in the discussion. 5) Regarding the presentation of economic impacts, I find it useful to show not only the total cost but also relative impacts, such as the cost per worker (e.g., relative changes in income per worker or USD lost per worker due to heat stress). One could also show the cost per worker in comparison to the income per worker in different regions to reveal potential impacts on income inequality and distribution across regions. Fig. 4 shows that in some regions, the annual economic loss could be larger than 200%, while the labour productivity loss does not exceed 8%. The authors should better explain the calculation of economic losses. The calculated economic costs seem to be implausible large. 7) As far as I know, the Bernard & Pourmoghani's method requires an iterative solution, but the method for WBGT calculation, which is described in Supplementary Information, does not seem like an iterative method. Please clarify it. Do you use an original Bernard & Pourmoghani's method or its simplification? Also, please add the formula for calculation of the dewpoint temperature. 8) I would suggest that the manuscript should undergo extensive English revisions because there are many typos and errors.

Responses to the reviewers' comments
Reviewer #1: In this study, the authors find out inequality labor loss risk, the economic cost of urban warming, cost at a sub-urban scale and the author is highly appreciated to focus on a key issue that is rarely studied. However, in its current form, the paper has some major shortcomings. Language editing is a must and the grammar and punctuations need to be edited for corrections throughout the manuscript. The manuscript requires major revisions with queries indicated that need to be addressed. The manuscript could make a worthy addition to the existing knowledge if the manuscript is revamped and written clearly, as it now lacks clarity and focuses on few areas.
1. Abstract: Mention the type of study and add study design and study period, method of data collection. Add more points on the methodology of the study so that it can be replicated by other researchers for their countries.

Response:
Information on the study design, study period and methods of data collection has been added to the abstract as basic research background. Please see the new abstract below: "We combine hourly heat stress data modeled with a regional climate model with exposureresponse functions between heat exposure and labor productivity obtained from related studies to investigate this possibility for 231 Chinese cities. We find that future urban heat stress is projected to cause an increase in labor losses of more than 0.20% total account GDP per year in the 2050s relative to the 2010s, with the largest increase projected to occur in mid-latitude coastal areas, and that some lower-paid sectors are disproportionally impacted."

Change the reference formatting in lines 64 and 65.
Response: The references were prepared using Endnote, and the reference formats have been checked and revised as appropriate throughout the manuscript. accordingly. Try to modify slightly to avoid redundancy as it is already apparent.

Response:
We have avoided redundancy according to the suggestion. In the Conclusion section, L78, 132, 143, 167 and 205 have been revised to avoid the start with the description of the tables or figures.

Mention the future CC Projection model name in the methodology section, even though it has been mentioned in the supplementary file.
Response: The projection model name has been added accordingly. The methodology section has been revised (L341-344). Please see the new contents below: "For the initial boundary conditions under the RCP 6.0 scenario, we used the global biascorrected dataset outputs from the National Center for Atmospheric Research (NCAR) community earth system model, which was used in phase 5 of the Coupled Model Intercomparison Project (CMIP5)." 7. The language part has to be corrected throughout the manuscript. Use free English editing software, which is available online. For e.g "Grammarly" to edit the language part. Response: In addition to using "Grammarly", we have had the revised manuscript edited for English language by American Journal Experts.
8. Line 89, 154 "strategies could reduce the average Tarea by over 0.20 °C in most of the 8 urban agglomerations" give spacing to "Tarea". Follow the uniformity throughout the manuscript.

Response:
The nonuniformity problem has been corrected here and throughout the manuscript; please see L102.

Reference number 21 is in line 119 but reference number 20 is in line 97. Be cautious while
citing the supporting articles.

Response:
The references were prepared using Endnote, and the reference formats have been checked and revised as appropriate throughout the manuscript.
10. As mentioned in the abstract "We combine hourly heat stress data with exposure-response functions between heat exposure and labor productivity to answer these questions for different cities in China under future climate change". I presume that you had used the formula to see the exposureresponse functions. Were any questionnaires used to collect this data? Kindly clarify.

Response:
In the revised version, we have described the use of only one type of epidemiological exposureresponse function without additional questionnaires as one of the limitations of the study; please see L284-286.
"we applied one type of epidemiological ERF, employing no additional questionnaires. The relationship between heat conditions and work loss could vary across the region. Therefore, more studies about the exposure-response relationship are needed." 11. Mention the type of statistical tool used for the analysis. 12. The citations in this manuscript should be in superscript format. Try to change all the references accordingly as per the guidelines.

Response:
As noted above, the references were prepared using Endnote, and the reference formats have been checked and revised as appropriate throughout the manuscript.

Add full forms in the table caption rather than the abbreviations
Response: The full forms have been added to the table captions accordingly.
14. Figure 1: try to represent the temperature variation in the map which has boundaries/lines for each province mentioned in the article rather than the national boundary. So that the reader can interpret the results easily.

Response:
The province boundaries have been added to  15. How the work intensity has been classified, mention the standard for that.

Response:
The work intensity was classified based on the following studies: To clarify this, we have cited these two references in the Methods section; please see L409.

Add the note for figure 4 and include the full form for all locations.
Response: The full name of all locations has been added to 18. Mention the full form of SSPs in line 208 as it is the first phrase that has an abbreviation (SSPs).

Response:
We have defined SSPs after its first appearance in the manuscript.

Response:
We added information about the study design and approach at the beginning of the Methods section. Please see L282 to L285. "By using a regional climate model, we aimed to quantify the potential influences of the interactions among climate change and potential urban adaptation strategies (installation of green roofs, cool walls, and cool ground surfaces) on future urban warming.
Based on a widely used ERF 17, 18 , we assessed the economic costs to urban residents induced by urban heat." Regarding the inclusion and exclusion criteria of the selected cities, the chosen cities were selected because their urban land area is larger than a certain size (10 square kilometers). We have added this information to the Methods section; please see L446-447. "The study involved a total of 18 sectors in 231 cities across mainland China. The cities were selected based on the criterion of an urban land area above a certain size (10 square kilometers)." The term 'urban canopy model' has been capitalized on L314. Response:

Conclusion
All of this information was included in the cover letter, and we cannot show this information due to the double-blind review process.

The manuscript estimates the heat-induced reductions in labor capacity for urban areas in
China under future climate change using a regional climate model. Only one climate change scenario was considered using RCP 6.0 and SSP2.

Response:
We thank the reviewer's comments. Base on the reviewer's suggestion, we have added another scenario, RCP 8.5 in the revised manuscript, and have compared our results with our previous results to make our results and discussion more reliable.
We have also compared the labor and economic losses under the two different climate change scenarios. The results show that the economic losses under the two different scenarios are comparable but that the spatial pattern differs between these two scenarios due to the different warming distributions.

Response:
We think there are three reasons for this obvious difference: a. We did not include all regions or all cities across China. Instead, we selected only 231 out of 371 cities in China, which were cities with a total urban land use area larger than a certain size (10 square kilometers).
b. Within these 231 selected cities, we accounted only for urban residents, not the entire population. In total, we accounted for 196 million exposed urban residents out of 1.4 billion people across China. We can do a simple calculation here. Taking China's per capita GDP in 2015 as a reference ($8066), the total GDP of 196 million people is 1580.9 billion U.S. dollars. So, the estimated annual economic loss (3 billion) account about 0.19% of total GDP from these exposed population. This result is consisted with the changes in regional GDP across East Asian in 2050 from Orlov et al. 2020 in Global Env. Change. For this, we have added more precise calculation process and results in the revised manuscript.
c. Our study does not cover all industries; rather, it covers 16 sectors involving urban economic activity. Other sectors, such as agriculture and forestry, are not considered. Therefore, a significant difference is expected between the calculated sum and the nationwide total.
To clarify this issue in the revised manuscript, first, we have explained that we did not account for all the cities but rather selected cities with a certain area of urban land use (L446-447). Second, we have clarified that this study only accounts for urban residents out of the entire population, as one of our aims is to assess the potential influence of urban adaptation (L407-409). Third, we have revised the presentation of our calculation. For comparison, we calculated the annual total GDPs from all selected urban grids in these 231 cities. The yearly total GDPs were obtained via two steps: 1. Calculate the accounted GDP for each city by multiplying the total urban residents within all selected urban grids by the GDP per capita of the corresponding city. 2. Sum the GDP values of each city to obtain the total accounted GDP (L447-451). Then, we calculated the changes in total GDP declines under RCP6.0 and RCP8.5 compared to the baseline scenario. As shown in Fig. 4, the results are comparable to results across East Asia reported in the cited study, i.e., Orlov et al. 2020 in Global Env. Change.
3.Also, the potential of urban adaptation strategies, which are assessed in the study, need to be put into perspective. The study reports that they "could offset more than 0.15 billion" (I assume USD, there was no unit reported). The study again does not discuss the relative magnitude of this major result except for Fig 4b. The adaptation strategies investigated could only offset 5% (or 10% as the 0.15 billion reported in the abstract contradicts Table S2 and Fig 4b) of the economic damage, despite the highly ambitious implementation rates. For instance, the green roof scenario assumes an adaptation rate of 80% across mainland China and the cool-wall scenarios assumes a reflective coating for all wall surfaces in urban areas. The scope and ambition of these scenarios are doubtful, and still produce only a very small impact. Why?
As a reader, one is left wondering why the authors did not focus on this part of the story:

1) Heat stress in Chinese major urban areas causes "just" 3 billion USD in annual economic
losses derived from reductions in labor capacity.

2) Very ambitious adaptation strategies only help to reduce this loss in worker productivity by 5 (or 10%). A possible or even likely reason is that the authors did not reflect about their results this way. If true, this substantially erodes my trust in how rigidly and carefully the analysis has been conducted and interpreted.
Response: With respect the first question, we have provided a detailed answer in the last issue.
Regarding the second question, first, we double-checked our assessment results to ensure that there were no calculation errors in the manuscript. As the reviewer points out, there were some unit and data inconsistency problems in the previous version. We have corrected these problems for the different future scenarios. Please see Table S1. Furthermore, we have compared our modeling results with results obtained with the same WRF input and settings, including Krayenhoff, E. Scott, et al. "Diurnal interaction between urban expansion, climate change and adaptation in US cities." Nature Climate Change 8.12 (2018): 1097-1103 can roll back warming of emerging megapolitan regions." Proceedings of the National Academy of Sciences 111.8 (2014Sciences 111.8 ( ): 2909Sciences 111.8 ( -2914. We find that the cooling impacts of the adaptation strategies in this study are comparable to those reported in these studies and that the cooling strategies could reduce the average Tarea by over 0.20 °C while higher average warming of 1-2 °C persists. That is, although these strategies can reduce heat pressure, with a cooling in excess of 1.5 °C at the peak of solar radiation, a gap remains relative to the average warming. To clarify this issue, we have added content to the discussion section; please see L271 to 276. Furthermore, regarding the cost of application of these adaptation measures, we calculated the additional cost of applying green roofs in the selected urban areas, as the installation of green roofs costs the most among the three selected measures. We calculated the total application cost for all of these cities, which reaches 18.3 billion US dollars. Based on the annual benefits of green roofs, it may take approximately 87 years to recover all the application costs. Many studies have indicated that these building measures can also cool indoor areas, thereby reducing the total energy use and daily costs during the warmest months. Overall, we think the benefit exceeds the application cost in the long run. 4. There are major results that are unrecognized (or unreflected), which seem to be in conflict with literature and which may point to a flaw in the data analysis and interpretation. Also, the results are taken from just one climate change scenario, without a sensitivity analysis and the adaptation strategy scenarios are highly unrealistic.

Moreover, crucial aspects are not covered by the analysis. What is the cost-effectiveness of the adaptation measures? The authors seem to implicitly assume no cost, but that is of course not the case. How will technology and the economy react towards increases in heat-stress? What is the
sensitivity of results coming from the model's and scenario's parameters and assumptions? I therefore doubt that this is a rigid analysis, which probably requires more than a major revision.

Response
As we noted previously, we have fully considered the comments of the reviewer. The problems can be divided into three areas: 1. the single scenario setting, 2. the significant different total value, and 3. the cost. We have addressed these issues in detail in the last three issues.

General remarks: Please use a professional English language editing service. The manuscript is difficult to understand in its content, which to a large part is owed to poor language and syntax. The reader is often left guessing what the authors' intended to say. This makes it quite
difficult to understand the details of the analysis. There are also many typos, which can be easily avoided using spell check.

Response:
We have had the revised manuscript edited for English language by American Journal Experts.

Remaining major comments: The study's main contribution, beside its empirical findings, is
on how in detail the economic structure and population residence is accounted for when analyzing the labor induced heat stress. This is indeed a valuable contribution, but quite technical and not well highlighted. It also does not seem to have a major impact on the results (the % bias it corrects is not explicitly reported, it seems to be less than 10%).

Response:
As we noted above, we have double-checked our assessment results to ensure that there are no calculation errors in the manuscript.
To highlight this part of the results, we have revised the discussion section, as described in our responses to comments. 2 and 3.

China may certainly be a very exciting case study, but the reader does not really see why.
This could be much better motivated at the beginning (e.g., X% of the world's urban population live in China).

Response:
Some reasons why we selected China as our case study and related information have been added accordingly on L33-36 of the Introduction section.
"This is especially true in China, where more than 2% of the world's urban population currently live and where more than 70% of the projected population increase will occur in urban areas 7.
Therefore, taking cities in China as examples can show the significant impact of continued urbanization on urban warming in the future."

The study does not discuss or consider how cost-effective the adaptation strategies could be.
This is a major shortcoming, but could be rectified.

Response:
We have added content to the discussion section, as described in our response to comment 3.

I appreciate that the authors try to understand how heat-stress impacts different economic
activities at the (sub-)urban level instead of the regional or even national scale. Yet, it remains unclear how the authors estimated the location of the economic activities and how exactly they have treated the seasonality of workloads (e.g., construction workers may not work during the midday heat). Also, such a contribution (while definitely valuable) may not be sufficient for a consideration in a top multi-disciplinary journal. This is also backed by the rather simplistic consideration of other dimensions (economic adjustments are absent, reliance on one SSP, RCP and exposure function, etc).
The economic estimation is very simple and static. There is no endogenous adaptive behavior.
Also, no changes in the intensity of the individual economic activities (or even technological improvement). However, the Chines economy is growing very rapidly and technologies are evolving, so is it reasonable to still assume the same economic structure of today?

Response:
First, we estimated the location of the economic activities. The assessment assumption of our study is based on the latest related studies, such as Zhang Yuqiang and Shindell Drew T. Costs from labor losses due to extreme heat in the USA attributable to climate change. Climatic Change 164, 1-18 (2021). In addition, we used a regional climate model to investigate the impact within the urban area.
Second, regarding future economic structures, our study covered 231 cities in total, and the economic structure of each city can be expected to change in a diverse manner in the future. Existing scenario datasets are unable to support us in completing these detailed assessments, as suggested by the reviewer.
However, we do not assume that economic development across cities will remain unchanged in the future. Our assessment is based mainly on the calculated changes in population exposure. To reflect the changes under future scenarios and the potential impact of future urbanization in China, we used the changes in the population distribution. As we mention in the manuscript, the urbanization rate is projected to grow in the future. In this process, the populations of some large cities will continue to increase while those of some small cities will decrease. Therefore, based on the projected population distribution, we can observe the development of urbanization and economic conditions across different cities, although not the detailed economic structure.
As this point was not discussed in the last version of the manuscript, we have added content addressing this issue in the Discussion section. Additionally, we agree that this is a limitation of our study and have added it to our description of the study limitations.
10. The discussion of how the analysis accounts for changes in the urban population are very difficult to comprehend. I think the authors at least once confuse "urbanization rate" with "share of total population living in urban areas" -see line 236.

Response:
According to the suggestion, we have revised the related content; please see L327-329.
"Regarding urbanization, the percentage of the total population of China that lives in urban areas is 60.9%. The Chinese government expects it to increase to 65.5% by 2030." Minor comments: 11. Defining transportation, manufacturing, etc. as medium work-intensity may overestimate heat-stress (also many of these activities can or are done in AC environments -especially until 2050).

Response:
We adopted this type of calculation method after referring to the relevant literature. Currently, there is no research quantifying how the work environments of these sectors will change in the future. Our research hypothesis is more comprehensive than those of other studies that do not consider AC coverage.
We agree this should be a limitation for this study, we have addressed this issue as a limitation in the Discussion section; please see L281 to 283. "We did not consider how AC coverage or working conditions of these sectors might change in the future, which may have led to some overestimate or underestimate, especially for types of medium-intensity work. Thus, we were unable to precisely estimate the impacts of future economic development other than future population change and changes in working conditions on urban labor loss." 2. I think the term "labor support" is not helpful for an interdisciplinary audience. I would also recommend to speak of "losses in labor capacity or productivity" than just "labor loss".
"The expected significant impacts of this increased heat exposure on human health have been studied in detail, while the impact on labor capacity or productivity due to outdoor heat remains unknown. The present study suggests that heat exposure among urban residents will significantly impact labor capacity or productivity across cities in China." 3. There are many typos, unclear sentences, strange formatting, etc. -this gives the reader a sloppy impression.

Response:
We have had the revised manuscript edited for English language by American Journal Experts.

Reviewer #3:
This study assesses the heat-induced impacts on labour productivity and associated economic costs under different adaptation measures at a sub-urban scale. The results from this impact assessment provide some novel and useful insights. Especially, I find interesting the combination of a regional climate model with an urban canopy model to investigate the effectiveness of different adaptation options in urban areas. I have a few questions and improvement suggestions:

Response:
In the revised manuscript, the comments of the reviewer have been fully considered. We have added another climate change scenario, RCP 8.5, and compared the results with our previous results, and our results and discussion are now more reliable.
We also compared the labor and following economic losses between the two climate change scenarios. The results show that the economic losses under the two different scenarios are comparable but that the spatial pattern differs between these two scenarios due to the different warming patterns.

I am a bit surprised that the economic cost from construction is so large compared to other
sectors. I understand that construction is more labour-intensive and requires working outdoors, but I guess much more people are working in manufacturing and services. So, I would expect that the total cost of heat-induced impacts on labour productivity in manufacturing and services would be larger than in construction. On that regard, I think it would be useful to have a table or figure showing the sectoral wage rates and the number of people employed in each sector and region.
Also, please report the recent penetration of air conditioners by regions.

Response:
We think the reasons of the biggest cost from construction is not only because of the high labor density, but also the completely exposure to the high temperature conditions without any coverage of air conditioning. Based on the assumption: represents the total number of residents of each urban grid from city j employed by sector i who are not protected by AC. represents the penetration rate of AC in city j for all sectors except the construction sector, as it mainly entails outdoor work. So, the total exposed population from construction would be higher than other industries, which leads to a larger total cost of heat-induced impacts on construction. It is consistent with the related study: Orlov Anton, Sillmann Jana, Aunan Kristin, Kjellstrom Tord, Aaheim Asbjørn. Economic costs of heat-induced reductions in worker productivity due to global warming. Global Environmental Change 63, 102087 (2020).
Furthermore, we have added the Table S3, which details the sector wage rates, the ratio of people employed in each sector and the recent penetration of air conditioners by region.
3. I would expect that the economic structure will change over time (i.e., more income will be generated from the service sector compared to manufacturing and construction). However, it seems that a structural change in the regional economies is not implemented in this study. In this case, this should be mentioned as a limitation.

Response:
Our assessment is based mainly on the calculated changes in population exposure. To reflect the changes under future scenarios and the potential impact of future urbanization in China, we used the changes in the population distribution. As we mention in the manuscript, the urbanization rate is projected to grow in the future. In this process, the populations of some large cities will continue to increase while those of some small cities will decrease. Therefore, based on the projected population distribution, we can observe the development of urbanization and economic conditions across different cities, although not the detailed economic structure.
We agree that this is a limitation of our study, and we have discussed it as such in the revised manuscript (L277-281). Response:

Also
Discussion of these adaptation measures has been added accordingly. Please see L274 to 276.

5.
Regarding the presentation of economic impacts, I find it useful to show not only the total cost but also relative impacts, such as the cost per worker (e.g., relative changes in income per worker or USD lost per worker due to heat stress). One could also show the cost per worker in comparison to the income per worker in different regions to reveal potential impacts on income inequality and distribution across regions.

Response:
We agree with the opinions of experts, but in consideration of suggestions by Reviewer 2 and to facilitate comparisons with related studies, we consider it preferable to calculate the changes in total GDP loss under RCP6.0 and RCP8.5 with SSP2 compared to the baseline scenario, as shown in Fig. 4.
6. Fig. 4 shows that in some regions, the annual economic loss could be larger than 200%, while the labour productivity loss does not exceed 8%. The authors should better explain the calculation of economic losses. The calculated economic costs seem to be implausible large. Response: As we described in our response to the previous comment, to clarify the results according to the reviewer's suggestion, we calculated the changes in total GDP loss under RCP6.0 and RCP8.5 with SSP2 compared to the baseline scenario, as shown in the revised Fig. 4. In addition, we compared these results to those of related research. We found that the total GDP account is consistent with that reported in a related study.
7. As far as I know, the Bernard & Pourmoghani's method requires an iterative solution, but the method for WBGT calculation, which is described in Supplementary Information We have had the revised manuscript edited for English language by American Journal Experts.