Clicks, Bricks, and Carbon: Digitalization’s Double-Edged Impact on Supply Chain Emissions Intensity

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper addresses a timely and important topic—the environmental implications of digitalization and e-commerce in supply chains. The empirical approach is employed, using panel data from European countries and advanced econometric techniques. The findings contribute to the literature by highlighting the conditional nature of digitalization’s impact on emissions intensity. However, several aspects could be strengthened.

1. The abstract is clear but could be more concise.
2. The introduction could be more engaging.
3. The literature review is comprehensive but could be better structured to build a clearer theoretical foundation for the hypotheses.
4. Hypotheses H2 and H3 are well-motivated, but H1 (“Higher digitalisation is associated with a decrease in emissions intensity”) seems overly simplistic given the mixed evidence cited.
5. The digitalization index is based on two indicators: ICT employment and fixed broadband subscriptions. While these are reasonable proxies, they may not fully capture the multidimensional nature of digitalization. Similarly, the e-commerce measure could be better described. How was it standardized? Is it based on turnover, penetration, or another metric?
6. The use of Driscoll-Kraay standard errors is appropriate for macro panels with cross-sectional dependence. However, the paper would benefit from a more detailed discussion of why this method was chosen over alternatives.
7. The GMM robustness check is a strength, but the results are not fully convincing due to weak instrument concerns.
8. The positive interaction between digitalization and e-commerce is intriguing but warrants more nuanced interpretation. Does this imply that digitalization amplifies the negative environmental effects of e-commerce? Or that e-commerce moderates the effect of digitalization?
9. The implications section is somewhat generic. Provide more specific recommendations for policymakers and managers.

Overall, this is a well-executed study with important findings. The conditional relationship between digitalization and emissions intensity is a valuable contribution to the literature. Therefore, I recommend a major revision.

Author Response

 

Dear Reviewer,

Thank you for the careful and constructive report. We revised the manuscript to address each point you raised and to improve clarity, methodological transparency, and practical relevance.

The abstract has been rewritten to state the research question, data window, main variables, identification approach, the associational nature of the estimates, and the principal empirical patterns, using neutral, academic language and passive voice. The introduction now frames the contribution around conditional effects of digitalisation, focusing on the interaction with e-commerce depth and the moderating role of the energy mix.

The literature review has been reorganised into three concise thematic subsections that mirror your request for a clearer synthesis: foundations in sustainable supply chain management and the digital capability lens; macro evidence, trade exposure and accounting perspectives; and operational mechanisms with energy-system context and governance. A short positioning paragraph at the end of the review links these streams to our hypotheses and design.

Measurement and preprocessing are documented with the precision you asked for. We added a compact “Data & variables” table listing definitions, sources, and model transformations. The digitalisation index is now defined as the first principal component of standardised enterprise IT adoption and fixed-broadband subscriptions per capita; we report oriented PC1 loadings and explained variance. The e-commerce measure is explicitly defined and its use justified; brief clarifications on renewables and other controls are provided. The rationale for retaining an unbalanced panel, constructing panel-safe lags, and mildly winsorising merchandise trade (1st/99th percentiles) is stated in a dedicated preprocessing paragraph.

Identification and inference are presented more transparently. The baseline uses two-way fixed effects with Driscoll–Kraay standard errors; we report and discuss core diagnostics (Breusch–Pagan, Jarque–Bera, Breusch–Godfrey, Pesaran CD, RESET, VIF). To probe persistence and potential endogeneity, we added parsimonious difference-GMM estimates with collapsed, depth-restricted instruments and we report AR(2) and over-identification tests. We also include complementary inference with cluster-by-country standard errors and country-specific linear trends.

Results are linked to interpretation in the way you requested. We report simple slopes of digitalisation evaluated at representative e-commerce quantiles and translate them into semi-elasticities, clarifying how the association varies across channel depth. Figures and tables have been streamlined; captions now state estimators, windows, and uncertainty conventions. A robustness table summarises the interaction coefficient across alternative outcomes, the extended window, clustered inference, country trends, and the GMM probes, alongside key diagnostics.

The managerial implications section has been rewritten as a concise table of observed patterns and recommended responses, followed by three short interpretive paragraphs that explain how consolidation, return prevention, packaging redesign, renewable procurement, and fleet electrification map to the empirical patterns. Language is neutral and academic. The limitations and future research section has been strengthened to acknowledge proxies, territorial accounting, unbalanced coverage, finite-sample identification limits, and the value of micro-data and quasi-experimental settings.

Formatting and MDPI compliance issues have been resolved: figure/table titles made unique and descriptive, abbreviations standardised, and notation harmonised. Throughout, we removed rhetorical markers and kept tone formal, clear, and human.

We appreciate your guidance. The revised manuscript implements the requested changes across abstract, literature review structure, data and preprocessing justifications, identification and diagnostics, robustness evidence, interpretation, managerial implications, and limitations. We hope the result meets your expectations and improves the paper’s clarity and usefulness.

Reviewer 2 Report

Comments and Suggestions for Authors

 

The article addresses a very pertinent issue in the current context of the digital and sustainable transition: the relationship between digitalization, e-commerce, and GHG emissions intensity in supply chains. The approach is relevant for both academic research and public policy and business management. However, the work appears more confirmatory than innovative. It lacks a clearer explanation of how the results advance the state of the art or challenge existing knowledge. The literature review is extensive and well-structured. The use of an unbalanced panel of 27 European countries (2014–2022), with two-way fixed-effects models and Driscoll-Kraay errors, demonstrates high empirical rigor. The hypotheses are well-defined and directly linked to the research question. The operationalization of the variables and the explanation of the econometric models are clear and detailed. The operationalization of the variables has some limitations: o The digitalization index, based on only two indicators (ICT adoption and broadband per capita), may be too limited to capture the complexity of the digital phenomenon. o The e-commerce variable lacks a more precise description and justification for its choice. o The absence of variables related to consumer behavior, environmental policy, or logistics infrastructure may introduce bias by omission. The discussion explores the underlying mechanisms and implications for businesses and policymakers (e.g., delivery density, urban policies, renewables integration). The suggestions are actionable and well-founded. The analysis acknowledges limitations regarding causality but does not propose stronger strategies to address it. Hypotheses H1 and H2 are rejected, which is addressed in the text, but the article does not sufficiently explore the theoretical implications of these rejections. Finally, the article acknowledges limitations (aggregate data, proxy measures, endogeneity issues) and suggests clear directions for future research, such as company-level microdata or the integration of consumption-based emissions. Some sections use long, imprecise sentences, or have grammatical structures that make them difficult to read. In particular, the introduction and discussion would benefit from linguistic revision for clarity and formality.

Author Response

 

Dear Reviewer,

Thank you for the detailed and constructive report. We have revised the manuscript to address each of your concerns and to improve clarity, contribution, and presentation.

We reframed the introduction to articulate more clearly what the paper adds. The opening now states the research question, the conditional perspective we test, and the two specific contributions: (i) separating general digital capability from the commercial channel by modeling the interaction between digitalization and e-commerce, and (ii) examining how the energy mix conditions this association. The final paragraph of the literature review now positions these contributions against prior work and explains why a multi-country panel is informative for this question.

We strengthened the operationalization of variables. Section “Data & variables” now includes a compact table with definitions, sources, and transformations, and a short methodological note on pre-processing (unbalanced panel by design, panel-safe lags, mild winsorization). For the digitalization index we report principal-component details (PC1 loadings for the two standardized inputs and explained variance) and orient the index to preserve interpretability. For e-commerce we clarify the choice of a business-side measure (enterprise online selling/ordering), explain comparability advantages over household or parcel-volume proxies, and note coverage trade-offs. These additions meet your request for a more precise description and justification.

The econometric design is now presented more transparently. We keep two-way fixed effects with Driscoll–Kraay inference and explain the choice relative to heteroskedasticity, serial correlation, and cross-sectional dependence in macro panels. We added a short “Identification & inference” paragraph that states assumptions and emphasizes the associational interpretation. Robustness checks were expanded and reported in a single summary table: alternative dependent variables, a longer window, country-specific linear trends, cluster-by-country standard errors, and parsimonious difference-GMM probes with instrument collapse and AR(2)/over-identification diagnostics. These changes respond to your request for clearer discussion of why the chosen method is appropriate and how sensitivity is evaluated.

To clarify mechanisms and implications, the discussion now connects the estimates to channel design (delivery density, return rates, packaging) and energy context, and provides a brief table of actionable levers for firms and policy (consolidation via lockers/PUDO, return prevention, packaging redesign, renewable sourcing and fleet electrification). This addresses your suggestion to offer more specific implications for business and policy.

We tightened the link between hypotheses and findings. H1–H3 are restated succinctly; the results section reports coefficients with uncertainty, simple slopes at representative e-commerce levels, and a joint Wald test for the “digital block,” followed by a short paragraph explaining which hypotheses are not supported and what that implies. We also clarified why some effects are imprecise under variants and how this relates to design choices rather than to a change of sign.

Limitations are expanded to reflect aggregate data, proxy choices, potential measurement error, production-based accounting, and finite-sample identification challenges; we indicate how future work with micro-data and quasi-experimental variation could strengthen causal claims, as you recommended.

Language and presentation were revised line-by-line to improve formality and readability. We shortened long sentences in the introduction and discussion, removed imprecise phrasing, harmonized notation and abbreviations, and ensured all figures and tables have unique, descriptive titles and consistent captions.

We hope these revisions address your comments and that the manuscript now presents its contribution and evidence more clearly.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

1. Consider adding a simplified conceptual diagram or table summarizing the model structure, variables, and their expected relationships.
2. The finding that digitalisation is associated with higher emissions intensity in high e-commerce contexts is central but could be better explained in economic and logistical terms.

Author Response

We thank the reviewer for this helpful suggestion. In the revised manuscript we have added a new Figure 1, titled “Conceptual framework linking digitalisation, e-commerce depth, renewable energy and GHG emissions intensity,” in Section 3 (Materials and Methods), immediately after the statement of H1–H3. The figure summarises the model structure, showing the direct effect of the digitalisation index, the moderating roles of e-commerce depth and the renewables share via the interaction terms, and the expected signs of these relationships. Figure numbering and cross-references have been updated accordingly.

We agree that this central finding requires a clearer economic and logistical interpretation. The Discussion section has therefore been expanded with two new paragraphs that describe how, in high e-commerce settings, digitalisation reduces search and ordering frictions, increases order frequency and basket fragmentation, and interacts with delivery windows, return policies and packaging practices to raise last-mile activity and reverse flows, potentially offsetting process efficiencies. The new text directly links these mechanisms to the estimated semi-elasticities at the median and upper quartile of e-commerce and contrasts them with contexts where e-commerce penetration is lower.