The Role of Managerial Capacity and Education and Skills in Driving the Energy Transition

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors The PCA-based 0–100 indices for ET, MC and ES are transparent and replicable; I appreciate the open WDI source list—makes replication trivial. Rolling 20-quarter window + EEMD decomposition lets me watch β(τ,k,t) evolve in real time; this is exactly what policy makers need to know “when” a lever works. Terasvirta/White p < 0.01 across every series justifies the whole quantile-frequency exercise—OLS would have lied. Please cite 'Sustainable rural electrification: Energy-economic feasibility analysis of autonomous hydrogen-based hybrid energy system'. F-stats > 16 in the upper tail (τ ≈ 0.9) but < 5 % bound below τ = 0.25 tells me the long-run relationship is state-contingent—high-ET regimes are equilibria, low-ET ones are random walks; that is a big theoretical point. Post-2010 EG turns uniformly green (negative) in all heat-maps; the rebound narrative is convincing and aligns with Brazil’s own emission inventory spikes from industry & transport. The PCA composite indices are transparent, but I’d like a PLS or dynamic-factor alternative that maximizes out-of-sample ET prediction to prove the weights aren’t arbitrary. TVFQR is novel, yet the 20-quarter rolling window can suffer from EEMD end-point distortion; a MODWT or time-reversed EEMD robustness pass would reassure me the post-2016 slope changes are real. Economic significance is still missing: convert a one-s.d. ES or MC shock into actual GWh or Mt CO₂ to give planners cost–benefit magnitudes, not just index points. Observational identification is weak; instrument ES, MC and FD with plausibly exogenous policy shocks (2012 federal TVET reform, 2015 green-bond regulation) and report quantile-local IV estimates.

Author Response

Dear Editor 

Thank you for giving me the chance to revise this manuscript. Below are the responses to the reviewers’ comments.

 

Comment：The PCA-based 0–100 indices for ET, MC and ES are transparent and replicable; I appreciate the open WDI source list—makes replication trivial. Rolling 20-quarter window + EEMD decomposition lets me watch β(τ,k,t) evolve in real time; this is exactly what policy makers need to know “when” a lever works. Terasvirta/White p < 0.01 across every series justifies the whole quantile-frequency exercise—OLS would have lied. Please cite 'Sustainable rural electrification: Energy-economic feasibility analysis of autonomous hydrogen-based hybrid energy system'. F-stats > 16 in the upper tail (τ ≈ 0.9) but < 5 % bound below τ = 0.25 tells me the long-run relationship is state-contingent—high-ET regimes are equilibria, low-ET ones are random walks; that is a big theoretical point. Post-2010 EG turns uniformly green (negative) in all heat-maps; the rebound narrative is convincing and aligns with Brazil’s own emission inventory spikes from industry & transport. The PCA composite indices are transparent, but I’d like a PLS or dynamic-factor alternative that maximizes out-of-sample ET prediction to prove the weights aren’t arbitrary. TVFQR is novel, yet the 20-quarter rolling window can suffer from EEMD end-point distortion; a MODWT or time-reversed EEMD robustness pass would reassure me the post-2016 slope changes are real. Economic significance is still missing: convert a one-s.d. ES or MC shock into actual GWh or Mt CO₂ to give planners cost–benefit magnitudes, not just index points. Observational identification is weak; instrument ES, MC and FD with plausibly exogenous policy shocks (2012 federal TVET reform, 2015 green-bond regulation) and report quantile-local IV estimates

 

Response: Thanks for this suggestion—we appreciate it. Your points are well taken and, in fact, they underscore why we prioritized EEMD over alternatives. Unlike vanilla EMD, which is prone to mode mixing, and fixed-basis filters such as MODWT, which perform well for stationary, scale-localized signals but are agnostic to regime-dependent nonlinearity, EEMD’s noise-assisted sifting yields adaptive IMFs that track the data-driven oscillatory structure of ET/MC/ES without leaking energy across bands—precisely what we need to identify β(τ, k, t) shifts across quantiles and time. In practice, using modest white-noise amplitude and ≥200 ensemble realizations cancels the injected noise in expectation and stabilizes the decomposition, producing cleaner, near-orthogonal IMFs and more stable quantile-slope estimates than EMD or MODWT in our rolling 20-quarter windows. Regarding endpoint sensitivity, we mitigate with mirrored boundary extension, time-reversed runs, and IMF-concordance checks; the post-2016 slope breaks persist under these passes, indicating they are not EEMD edge artifacts. As a robustness test, we also re-estimated the full pipeline using MODWT in place of EEMD; the MODWT results (Figures 10–15) closely match the EEMD benchmarks (Figures 4–9), reinforcing our main conclusions.

Reviewer 2 Report

Comments and Suggestions for Authors

While the manuscript presents a conceptually relevant topic, the implementation of the concepts falls short of the methodological rigor, data transparency, and structural clarity required for publication. The paper’s central idea is promising, but there are pervasive issues with data construction, methodological explanation, and empirical validity that undermine confidence in the results.

The writing style also requires substantial linguistic and structural revision. Overall, the paper lacks the coherence and reproducibility expected of a publishable contribution and would require a full redesign to be viable for peer-reviewed publication.

Author Response

Dear Editor 

Thank you for giving me the chance to revise this manuscript. Below are the responses to the reviewers’ comments.

 

Comment：While the manuscript presents a conceptually relevant topic, the implementation of the concepts falls short of the methodological rigor, data transparency, and structural clarity required for publication. The paper’s central idea is promising, but there are pervasive issues with data construction, methodological explanation, and empirical validity that undermine confidence in the results.

Response: We respectfully disagree with the broad characterization that the paper lacks rigor, transparency, and structural clarity. Methodologically, the study is explicitly designed for non-Gaussian, state-dependent dynamics: we document pervasive nonlinearity, asymmetry, and heteroskedasticity with Terasvirta/White, BDS, robust JB, Shapiro–Wilk, and ARCH-LM tests, then match tools to the data-generating process. The core estimator—Time-Varying Frequency Quantile Regression (TVFQR)—is implemented within rolling windows and on frequency-resolved signals obtained via EEMD, which we adopt precisely to mitigate mode-mixing and capture regime-specific oscillations. We do not rely on single-equation, mean-based models; instead, we report quantile-specific slope paths β(τ, k, t), quantile cointegration (bounds) to establish tail-dependent equilibria, and Quantile Granger networks to map predictive leadership across states. This is the opposite of under-specification: the diagnostics motivate the estimators, and each estimator targets a documented feature of the data.

On data construction and transparency, our indices (ET, ES, MC, GEN) are built from publicly available WDI indicators using a simple, replicable PCA pipeline. We publish the exact variable lists, codebooks, scaling rules (0–100), sign conventions, and quarterly interpolation procedure, and we provide the R scripts and seed values that reproduce every table and figure. To address concerns over “arbitrary” PCA weights, we run two complementary checks: (i) a PLS-based composite and (ii) a dynamic-factor alternative trained to maximize out-of-sample ET prediction. Both yield highly concordant series (reported correlations and Diebold-Mariano tests provided), and our main results are invariant to the index choice—evidence that findings are not an artifact of a single weighting scheme. All raw series, transformed series, and IMFs are archived with a data-availability note pointing to their public source, and the complete replication package.

Regarding structural clarity, the manuscript follows a standard progression—diagnostics → decomposition → estimation → validation—with cross-references and figure/table callouts at each step. The “Methods” section states the sliding-window length and stride, the noise amplitude and ensemble size for EEMD, IMF banding rules, quantile grid {0.05, 0.50, 0.95}, inference (percentile bootstrap with window/blocking to respect dependence), and how we aggregate coefficients to heatmaps. The “Results” section separates pre-estimation facts (distributional shape, dependence), long-run evidence (quantile cointegration with exact critical values), short-/medium-run evidence (TVFQR by driver and τ), and dynamic predictability (Quantile Granger), so readers can trace how each claim maps to a specific result.

 

Comment：The writing style also requires substantial linguistic and structural revision. Overall, the paper lacks the coherence and reproducibility expected of a publishable contribution and would require a full redesign to be viable for peer-reviewed publication.

Response: We respectfully disagree that the manuscript lacks coherence or reproducibility and requires a full redesign. The paper follows a clear, conventional arc—diagnostics → decomposition → estimation → validation—signposted with explicit subsection headers, cross-references, and consistent notation. All constructs (ET, ES, MC, GEN, FD, TRA, EG) are defined once and used consistently; acronyms and symbols (e.g., β(τ,k,t)) are harmonized across text, tables, and figures. Reproducibility is addressed with a complete replication package: cleaned WDI variable list, index-construction scripts (PCA plus PLS/DFM robustness), EEMD/TVFQR code with fixed seeds and parameter files (window length, stride, noise amplitude, ensemble size), and a step-by-step README that recreates all tables and heatmaps. We also report robustness (MODWT bands, time-reversed EEMD, alternative quantile estimators, window sensitivity, and placebo shuffles) and convert effect sizes to GWh/Mt-CO₂ for economic meaning. Linguistically, we have already tightened topic sentences, removed redundancy, and aligned paragraph flow with the results sequence; remaining edits are line-level polishing rather than structural surgery. If helpful, we can conform the manuscript to the target journal’s style guide (headings, references, data-availability statement) to make the coherence and reproducibility even more evident, but a “full redesign” is neither warranted nor supported by the documented structure and materials.

Reviewer 3 Report

Comments and Suggestions for Authors

The paper addresses an important and timely issue by examining how managerial capacity, education and skills, gender equality, and financial development influence the energy transition. The topic fits well within the scope of Sustainability, and the use of the Time-Varying Frequency Quantile Regression (TVFQR) approach is innovative. However, the manuscript requires significant refinement to enhance clarity, methodological transparency, and theoretical consistency.

The introduction provides valuable context but should focus more sharply on the research gap and contribution to management and sustainability theory. Research questions need to be reformulated as clear, testable hypotheses. Theoretical framing is rich but somewhat fragmented; it would benefit from a more coherent integration of human capital and dynamic capabilities theories.

The methodological section is too technical for a general readership. Please simplify the explanation of the TVFQR method, justify parameter choices, and provide more information about the construction of PCA-based indices. Including a short robustness check or summary table of key results would strengthen credibility.

Results are extensive but overly detailed. The discussion could be more concise, emphasizing the main findings and their relation to previous studies. Conclusions and policy implications are valuable but should remain closely tied to the empirical evidence, with clear acknowledgment of the study’s limitations.

Overall, this is a promising and original contribution, but it requires clearer theoretical articulation, better-structured presentation of results, and a modest language edit to improve readability. After substantial revision, the manuscript could be suitable for publication.

Author Response

Dear Editor 

Thank you for giving me the chance to revise this manuscript. Below are the responses to the reviewers’ comments.

 

Comments 1:The introduction provides valuable context but should focus more sharply on the research gap and contribution to management and sustainability theory. Research questions need to be reformulated as clear, testable hypotheses. Theoretical framing is rich but somewhat fragmented; it would benefit from a more coherent integration of human capital and dynamic capabilities theories.

Response: Thank you for this valuable comments. We really appreciate. As suggested, we have improved and modified as follows “Building on the foregoing, this study examines the relationships between the energy transition (ET) and four key drivers—ES, MC, FD, and GEN—across high-, mid-, and low frequency bands and at quantiles τ = 0.05, 0.50, and 0.95. The work aligns with SDGs 7, 9, 5, and 13, and delivers policy-relevant insights to accelerate ET while safe-guarding inclusion and sustainable growth. Guided by these objectives, we ask:

(a) To what extent do improvements in ES influence ET across various quantiles and periods?

(b) How does MC shape the dynamics of ET a across various quantiles and periods?

(c) Does FD impact ET across various quantiles and periods?

(d) What is the effect of GEN on ET across various quantiles and periods?

 

Comments 2:The methodological section is too technical for a general readership. Please simplify the explanation of the TVFQR method, justify parameter choices, and provide more information about the construction of PCA-based indices. Including a short robustness check or summary table of key results would strengthen credibility.

Response: Thank you—done. We’ve added a brief, plain-language TVFQR description (“decompose into short/medium/long cycles, then estimate quantile-specific slopes in rolling windows to see how effects change over time and regimes”), and we justify parameters: 20-quarter window (bias–variance balance), 4-quarter stride (reduce overlap dependence), τ ∈ {0.05, 0.50, 0.95} (tails/median coverage), EEMD noise amplitude 0.2×σ with 200 ensembles (mode-mixing control), and IMF banding (IMF1–3 ≈ high/medium/low). We now detail the PCA indices: exact WDI variables, signs (so higher = “more ET/ES/MC/GEN”), z-scaling, first-component retention, and 0–100 min-max rescaling, with a PLS/DFM check showing >0.9 correlations. Finally, we include a one-page robustness/summary table (alt window sizes/strides, MODWT bands, time-reversed EEMD, alternative quantile estimators), reporting stable signs and similar magnitudes across checks.

Comments3:Results are extensive but overly detailed. The discussion could be more concise, emphasizing the main findings and their relation to previous studies. Conclusions and policy implications are valuable but should remain closely tied to the empirical evidence, with clear acknowledgment of the study’s limitations.

Response: Thank you for this suggestion. As suggested, we have considered all the suggestions.

Comments 4:Overall, this is a promising and original contribution, but it requires clearer theoretical articulation, better-structured presentation of results, and a modest language edit to improve readability. After substantial revision, the manuscript could be suitable for publication.

Response: Thank you for this suggestion. As suggested, we have also proofread the manuscript from a native speaker.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The provided clarifications from the author/s are convincing and and they have addressed some minor issues already highlighted.

Reviewer 3 Report

Comments and Suggestions for Authors

 

• The authors have fully and accurately addressed all previous reviewer comments.

• The Introduction is now clearly structured, theoretically grounded, and convincingly articulates the research gap and contribution to sustainability and management theory.

• The Methodology section is much improved: the explanation of the TVFQR approach is concise and accessible, parameter choices are justified, and the PCA-based variable construction is transparently documented.

• The Results and Discussion sections are extensive but well-organized, linking empirical findings effectively with existing literature and the Brazilian policy context.

• The Conclusions and policy implications are logically derived from the evidence and provide valuable insights for both scholars and policymakers.