Cloud Computing and Green Total Factor Productivity in Urban China: Evidence from a Spatial Difference-in-Differences Approach

Comments 1:  Although the paper emphasizes “the impact of cloud computing on green total factor productivity (GTFP),” its innovative contributions are insufficiently articulated compared to existing research on the digital economy, artificial intelligence, and big data. We recommend more clearly distinguishing this study's differences and unique contributions from prior literature in both the introduction and conclusion.

Response 1: We thank the reviewers for their constructive comments. In response, we have made the following improvements to refining the contributions of this paper:

(1)We have further summarized the unique contributions of this paper in the Introduction section, specifically in Section 1, on page 3, lines 120–145.

(2) We have reiterated the key contributions of this paper in the Conclusion section, on page 33, lines 954–968.

Comments 2: We suggest presenting the causal relationships of Hypotheses 1–3 through a concise mechanism diagram rather than through a lengthy literature review.

Response 2: We would like to thank the reviewers for pointing out the issues. We have already prepared a framework diagram illustrating the interrelationships among the hypotheses in Figure 1 (in Section 2.3, on page 6, line 249 - 257).

Comments 3: The paper proposes three hypotheses (H1–H3), but the results section lacks a tight correspondence with these hypotheses. It is recommended that the discussion section explicitly address each hypothesis, clearly indicating which are fully supported and which are only partially supported. 

Response 3: We thank the reviewers for their comments on the discussion of hypotheses in this paper. To provide a more detailed and thorough discussion of the hypotheses in the empirical analysis, we have:

(1) discussed whether Hypothesis 1 is supported in Section 4.3, on page 15, lines 508–511;

(2) provided a more detailed discussion of whether Hypothesis 2 is supported in Section 5.1, on page 24, lines 693–697;

(3) given an even more detailed discussion of Hypothesis 3 in Section 5.1, on page 25, lines 712–718.

Overall, based on our analysis, we believe that the empirical results of this paper provide sufficient support for Hypotheses 1–3.

Comments 4: Substituting “total electricity consumption” for energy input, though precedent exists, lacks sufficient justification. We recommend detailing the limitations of this metric and adding comparisons with other energy consumption data or sensitivity tests. 

Response 4: We thank the reviewers for pointing out the limitations. To address the limitations related to the total societal electricity consumption and conduct sensitivity analysis:

(1) We have pointed out the limitation of using electricity consumption as a proxy for energy input in Section 4.6.7, on page 23, lines 655–660.

(2) We have estimated the total energy consumption at the prefecture-level cities by following the method of Yang et al.(2023) [1] and Chand et al.(2009) [2]; meanwhile, following the approach of Liu et al.(2025) [3] and Xia and Xu(2020) [4], we have recalculated GTFP without considering energy input to conduct sensitivity analysis. The results in Section 4.6.7, on page 23, line 661 - 677, show that our conclusions remain robust under different energy input considerations.

Reference

1. Yang G.; Wang H.; Fan H.; Yue Z. Carbon Reduction Effect of Digital Economy:Theoretical Analysis and Empirical Evidence. China Industrial Economics 2023, 80–98, doi:10.19581/j.cnki.ciejournal.2023.05.005.

2. Chand, T.R.K.; Badarinath, K.V.S.; Elvidge, C.D.; Tuttle, B.T. Spatial Characterization of Electrical Power Consumption Patterns over India Using Temporal DMSP‐OLS Night‐time Satellite Data. International Journal of Remote Sensing 2009, 30, 647–661, doi:10.1080/01431160802345685.

3. Liu, H.; Fang, W.; Yuan, P.; Dong, X. How Does Climate Change Affect Green Total Factor Productivity? Climatic Change 2025, 178, 112, doi:10.1007/s10584-025-03945-0.

4. Xia, F.; Xu, J. Green Total Factor Productivity: A Re-Examination of Quality of Growth for Provinces in China. China Economic Review 2020, 62, 101454, doi:10.1016/j.chieco.2020.101454.

Comments 5: The paper reports numerous significant coefficients but inadequately explains their economic implications (e.g., how coefficient magnitudes translate to actual GTFP improvements).

Response 5: We thank you for your supplementary comments on the interpretation of the empirical results in this paper. We have conducted a specific discussion on the economic significance of the coefficients in Section 4.3, on page 15, lines 495–506.

Comments 6: The article lacks in-depth discussion on regional policy implications. Explore how to mitigate these negative spatial spillovers at the policy level.

Response 6:  We fully agree with the reviewer’s point regarding the need to further mitigate the cloud computing siphon effect.

Following the idea you provided, to address this effect on urban GTFP, we have specifically discussed in Section 5.4, on pages 29-32, lines 847–947, the role of different regional policies in mitigating the cloud computing siphon effect.

We find that regional integration policies, administrative monopoly policies, and trade liberalization policies can mitigate its negative siphon effect.

Comments 7: Further consider dimensions such as economic development levels and digital infrastructure levels to enhance the universality of conclusions.

Response 7: We thank you for your supplementary comments.

First, to rule out the influence of urban economic development on the results, in Section 4.6.5 , page 21, lines 607–636, we further controlled for the natural logarithm of city GDP (lngdp) and the natural logarithm of per capita city GDP (lnper_gdp) in the regressions.

Second, the application of urban cloud computing relies on digital infrastructure: sound digital infrastructure improves data transmission capacity, enhances computing power, and facilitates the construction and promotion of urban cloud computing.Currently, many studies use the Broadband China Initiative as an exogenous shock to digital infrastructure. Accordingly, we also employ it as a proxy for digital infrastructure construction. Meanwhile, the Chinese government has launched the Information Benefiting the People Project—a strategic initiative to drive leapfrog development in people’s livelihood areas through information technology. During the project’s implementation, it has improved information infrastructure, further addressing interconnection challenges in livelihood domains such as health, medical care, elderly care, employment, and domestic services. Overall, we use the Broadband China Initiative (Dboard) and the Information Benefiting the People Project (Dinformation) as proxies for digital infrastructure.

After controlling for economic factors and digital infrastructure, our conclusions remain robust.

Comments 8:The policy recommendations section fails to align closely with the research findings. More targeted measures are needed on how to leverage cloud computing to promote green innovation and avoid competitive negative spillovers between neighboring cities. 

Response 8: We thank the reviewer for their comments on the policy recommendations in this paper. To better align the policy recommendations with the empirical analysis of this study, we have revised them in Section 6, on pages 34-35, lines 1019–1060.

Comments 9: Reference: Critical Success Factors for Green Port Transformation Using Digital Technology

Response 9: This article has provided a positive reference for the current study and has been cited in [69].

1. Summary

Dear Reviewers,

Thank you for your valuable comments, which have been instrumental in enhancing the quality of this paper.

Greetings to you all. We sincerely appreciate you taking the time out of your busy schedules to conduct a rigorous and thorough review of our manuscript. Thank you for your patient review and valuable suggestions regarding structural arrangement, format adjustments, and data explanations. Your professional advice and guidance have provided profound insights, helping us identify shortcomings in the paper, enhance its quality, refine research content, standardize empirical processes, and optimize language presentation. These contributions have played a crucial role in elevating the academic value of this work.

We greatly value your feedback. We have carefully read, thoroughly considered, and integrated all your valuable suggestions into the revised manuscript. In the updated version, we have addressed and refined each point raised, and on this basis, expanded, optimized, and enriched the research content. Below, we provide point-by-point responses to your comments and detailed descriptions of the corresponding revisions.

Once again, we sincerely thank you for your careful review and patient feedback.

First, to refine the academic English expression of this paper, we have conducted detailed linguistic revisions throughout the text. This was done by carefully proofreading and adjusting the manuscript, building on the author services provided by the Sustainability journal.

Second, we have provided the following explanations for the revisions based on your comments. Since some revisions are extensive, we have indicated the specific locations of the changes in our response and highlighted the revised content in yellow.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Is the content succinctly described and contextualized with respect to previous and present theoretical background and empirical research (if applicable) on the topic?

Yes

The relevant content will be revised in the subsequent response.

Are the research design, questions, hypotheses and methods clearly stated?

Yes

The relevant content will be revised in the subsequent response.

Are the arguments and discussion of findings coherent, balanced and compelling?

Can be improved

The relevant content will be revised in the subsequent response.

For empirical research, are the results clearly presented?

Yes

The relevant content will be revised in the subsequent response.

For empirical research, are the results clearly presented?

Yes

The relevant content will be revised in the subsequent response.

For empirical research, are the results clearly presented?

Yes

The relevant content will be revised in the subsequent response.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: Please check the size of the table

Response 1: We would like to thank the reviewers for pointing out the formatting issues. We have readjusted the size of all tables in accordance with the journal's formatting requirements.

Comments 2:While the use of a Difference-in-Differences (DID) framework suggests a quasi-natural experimental design, the abstract refers to extracting the "policy adoption timeline" via text analysis. It would be highly beneficial to briefly clarify which specific policy or event serves as the "treatment"—either in the abstract or in a footnote. For spatial DID models in particular, precise information on the timing and geographic scope of treatment adoption is critical for correct interpretation and replication.

Response 2: We sincerely appreciate the valuable comments provided by the reviewers. We have clarified in footnote [1] (page 10, Section 3.3.1) that this study uses the timing of the establishment of the first cloud computing service center in each city as the "treatment variable".

Further details are provided below:

This study uses Chinese prefecture-level cities as analytical units to examine the impact of cloud computing. There is variation in the timing of cloud computing adoption across cities, with no single uniform policy cutoff date. Our data collection reveals that each city has its own timeline for adopting cloud computing. As cloud computing continues to proliferate, this study defines the establishment of a city’s first cloud computing service center as the policy timing for cloud computing adoption, classifying cities that established such centers as the treatment group.

For example, Beijing established the Beijing Supercomputing Cloud Center  in 2011, a national-level information infrastructure platform co-built by the Beijing Municipal People’s Government and academic institutions for supercomputing and cloud services. We therefore set 2011 as the policy timing for Beijing’s cloud computing implementation. In Tianjin, the earliest cloud computing service center is the National Supercomputing Center Tianjin.  Approved in May 2009, it underwent installation and debugging in June 2011, became operational in November 2011, and officially launched commercial computing power services in 2012. We thus set 2012 as the policy timing for Tianjin’s cloud computing services. By compiling the establishment dates of the first cloud computing service centers in each prefecture-level city, we obtain the policy timing for cloud computing adoption in each city.

We sincerely appreciate your valuable suggestions, which have significantly contributed to improving the quality of this paper.

Comments 3: Please explain why do smaller cities exhibit a stronger siphon effect? (Perhaps due to lower baseline capacity or lack of complementary resources). Why do coastal regions generate positive spillovers? (Perhaps due to stronger industrial linkages and superior infrastructure).

Response 3: We would like to thank the reviewers for their analytical insights, and this paper provides the following explanations.

(1) Reasons why small-scale cities face a stronger siphon effect (See lines 744–767, page 26, Section 5.2):

   For small-scale cities, urban capital investment and infrastructure development are relatively backward. For instance, the average log capital stock is 16.12 in samll-scale cities, compared to 17.26 in larger-scale cities, highlighting their lower investment level.. In small-scale cities, capital further flows to large cities, thus exposing them to a more pronounced siphon effect. Meanwhile, in cities implementing cloud computing, the continuous improvement in industrial development and digitalization levels may trigger a siphon effect, where talent and capital flow toward these cloud computing cities. While the agglomeration effect of cloud computing enhances local GTFP, it may simultaneously suppress the green development of surrounding small-scale cities through factor drainage mechanisms. Cloud computing cities, equipped with advanced digital infrastructure and green innovation platforms, establish platforms that improve massive data processing capabilities and foster R&D collaboration [1,2], creating strong attractions for technical talent and digitally skilled labor. Small-scale cities, however, have low resistance to large cities and struggle to prevent the outflow of talent and resources. Second, as an emerging information technology critical to future urban economic development [3–5], cloud computing exhibits comparative advantages in green project financing, digital technology investment, and ecosystem modeling [6]. This makes it highly attractive to regional venture capital and industrial funds, diverting financial resources from green industry projects in surrounding small-scale cities. Consequently, small-scale cities struggle to advance infrastructure investments in clean energy substitution and circular economy parks, slowing improvements in resource utilization efficiency and the green transition process.Thus, the development of urban cloud computing may generate siphon effects that widen regional gaps in green development, exerting adverse impacts on the GTFP of neighboring small-scale cities.

(2) Reasons why coastal regions demonstrate positive spillover effects, and this paper offers the following explanations (See Section 5.2, page 27, lines 787–806, ):

The significant positive spillover of cloud computing in coastal regions on urban GTFP stems primarily from strengthened inter-regional cooperation. First, coastal regions (e.g., the Yangtze River Delta and Pearl River Delta) have mature industrial coordination mechanisms, enabling leading digital economy firms and neighboring cities to form rational division of labor chains [7,8]. Central cities focus on breakthroughs in core technologies such as algorithms and big data, while smaller cities undertake segments including smart environmental protection equipment manufacturing and industrial internet applications. Cross-city data interoperability optimizes production processes , reduceing resource misallocation and redundant construction [9,10]. Second, the interconnectivity of infrastructure such as ports and high-speed railways facilitates inter-regional cooperation [11]. Coastal regions, with more comprehensive transportation infrastructure networks, facilitate inter-regional collaboration and drive the positive technological spillover of cloud computing. Third, lower administrative monopolies can encourage technological exchange and regional cooperation [12]. Coastal regions exhibit higher marketization levels and lower administrative monopolies [13,14], enabling the rapid diffusion of green technologies within the region and avoiding the short-sighted "beggar-thy-neighbor" behavior. In summary, the stronger coordination and cooperation, lower administrative monopolies, and well-developed infrastructure in coastal regions collectively promote positive spillovers in GTFP.

Reference

Golightly, L.; Chang, V.; Xu, Q.A.; Gao, X.; Liu, B.S. Adoption of Cloud Computing as Innovation in the Organization. International Journal of Engineering Business Management 2022, 14, 18479790221093992, doi:10.1177/18479790221093992.

2. Lei, L.; Feng, H.; Ren, J. Artificial Intelligence, Human Capital and Firm-Level Total Factor Productivity. Finance Research Letters 2025, 107897, doi:10.1016/j.frl.2025.107897.

3. Chen, Y.; Zhang, R.; Lyu, J.; Ma, X. The Butterfly Effect of Cloud Computing on the Low-Carbon Economy. Technological Forecasting and Social Change 2024, 204, 123433, doi:10.1016/j.techfore.2024.123433.

4. Han, L.; Wojan, T.R.; Goetz, S.J. Cloud Computing and Rural Globalization: Evidence for the U.S. Nonfarm Economy. Telecommunications Policy 2024, 48, 102814, doi:10.1016/j.telpol.2024.102814.

5. Polyviou, A.; Venters, W.; Pouloudi, N. Distant but Close: Locational, Relational and Temporal Proximity in Cloud Computing Adoption. Journal of Information Technology 2024, 39, 71–93, doi:10.1177/02683962231186161.

6. Floerecke, S.; Lehner, F.; Schweikl, S. Cloud Computing Ecosystem Model: Evaluation and Role Clusters. Electron Markets 2021, 31, 923–943, doi:10.1007/s12525-020-00419-2.

7. Fan Jianyong Yangtze River Delta Integration, Regional Specialization, and the Spatial Relocation of Manufacturing. Journal of Management World 2004, 77–84, 96, doi:10.19744/j.cnki.11-1235/f.2004.11.011.

8. XiaoFeng C.; ZhaoFeng C. Level and Effect on Co -agglomeration of Producer Service and Manufacturing Industry:  Empirical Evidence from the Eastern Area of China. Finance and Trade Research 2014, 25, 49–57, doi:10.19337/j.cnki.34-1093/f.2014.02.007.

9. Xuesong M.; Dingliang W. The Internal Mechanism and Realization Path for the Modernization of Spatial Governance in Megacities from the Perspective of Deepening Coordinated Regional Development. Social Sciences in Yunnan 2025, 23–33.

10. Xiaolin W. The City of Traffic in the Digital Age:The Rise and Governance of the New City Form. Jiangsu Social Sciences 2022, 62–72, 242–243, doi:10.13858/j.cnki.cn32-1312/c.20220722.022.

11. Chengyang X.; Wang Minghui A Study on the Impact of Transportation Infrastructure on the Spatial Distribution of Industrial Activities. Journal of Management World 2020, 36, 52–64, 161, 65–66, doi:10.19744/j.cnki.11-1235/f.2020.0183.

12. Xiaojuan J.; Lijun M. Mainly Inner Circulation, Outer Circulation Empowerment and Higher Level Double Circulation:International Experience and Chinese Practice. Journal of Management World 2021, 37, 1–19, doi:10.19744/j.cnki.11-1235/f.2021.0001.

13. Liangchun Y.; Donghua Y. The Measurement of Local Administrative Monopoly Degree in China. Economic Research Journal 2009, 44, 119–131, doi:CNKI:SUN:JJYJ.0.2009-02-011.

14. Xianxiang L.; Susu W. Measurement of the Degree of Marketization of Factors, Decomposition of Regional Differences and Dynamic Evolution: Empirical Research Based on China’s Provincial Panel Data. South China Journal of Economics 2021, 37–63, doi:10.19592/j.cnki.scje.381608.

Comments 4: Line 667: The sentence "X is the control variables, and WX is the spatial term of the control variables" is repeated twice. Please remove the repetition. 

Response 4: We sincerely thank the reviewers for their careful review. We have removed redundant content and carefully proofread the rest of the manuscript.

Comments 5: The panel data spans from 2000 to 2023 (line 17, 676). Cloud computing adoption began in earnest much later than 2000. Briefly justify including the long pre-treatment period (2000-adoption year) in the paper, as a long pre-period aids in establishing the parallel trend assumption of the DID model

Response 5: We thank the reviewer for their comments. To ensure more robust results, first, we further use 2003, 2005, and 2008 as alternative start years to re-estimate the model. Second, we retain only 9, 7, and 5 pre-treatment periods for re-estimation. These results are presented in Section 4.6.6, pages 22-23, lines 637–653.

Further details are provided below:

The Difference-in-Differences (DID) model requires no significant differences between the treatment and control groups prior to policy implementation, meaning: where D=1 denotes the treatment group. To test this assumption, existing studies employ the event study method for parallel trends testing. Prior to the adoption of cloud computing, there were no significant differences in GTFP between the treatment and control groups. In the parallel trends test presented in Figure 5 of this study, we analyze the GTFP differences between the treatment and control groups in the 10 pre-policy periods before cloud computing adoption and find no significant differences between them.

The year 2000 predates the period of urban cloud computing adoption. The reasons for using 2000 as the sample starting year are as follows: First, using an earlier start year of 2000 allows for a longer pre-policy sample. A longer pre-treatment sample enables clearer observation of GTFP differences between treatment and control group cities. There is no unified view on the length of the pre-treatment sample; some studies use longer pre-treatment periods, such as Nathan Nunn and Nancy Qian (2011) [15] in their article in The Quarterly Journal of Economics, which includes 12 pre-treatment periods; Beck et al. (2022) [16] also have 10 pre-treatment periods in their study; and Lovenheim and Willén (2019) [17] have 11 pre-treatment periods. Thus, using 2000 as the sample start to obtain 10 pre-treatment periods is generally reasonable.

Second, parallel trends testing typically verifies the validity of the assumption from two dimensions. (1) Prior to the policy timing, the outcome variables of the treatment and control groups should show no significant level differences, which is reflected in the insignificant coefficients of the “pre-policy dummy variables” included in the model. (2) The control group and treatment group should maintain parallel growth paths in the pre-period, and their estimated values do not exhibit monotonic trends that are economically or statistically significant.

In this study, using 2000 as the sample start and analyzing 10 pre-treatment periods, we observe that the pre-policy coefficients cluster around 0 with no clear trend, further confirming the validity of our parallel trends assumption. Third, using 2000–2023 as the sample period allows better capture of the long-term effects of urban cloud computing on GTFP. Additionally, samples before 2000 suffer from changes in statistical methods and significant data missingness; thus, we use 2000 as the sample start from the perspective of data availability.

Overall, considering existing research practices, the validity of parallel trends, the long-term policy effects, and data availability, using 2000 as the sample start year is reasonable.

Reference

15. Nunn, N.; Qian, N. The Potato’s Contribution to Population and Urbanization: Evidence From A Historical Experiment. The Quarterly Journal of Economics 2011, 126, 593–650, doi:10.1093/qje/qjr009.

16. Beck, T.; Levine, R.; Levkov, A. Big Bad Banks? The Winners and Losers from Bank Deregulation in the United States. The Journal of Finance 2010, 65, 1637–1667, doi:10.1111/j.1540-6261.2010.01589.x.

17. Lovenheim, M.F.; Willén, A. The Long-Run Effects of Teacher Collective Bargaining. American Economic Journal: Economic Policy 2019, 11, 292–324, doi:10.1257/pol.20170570.