1. You claim to investigate “asymmetric impacts” and highlight nonlinear relationships, yet Table 7 shows remarkably consistent coefficient signs and only modest magnitude variations across all quantiles (CHE: 2.937 at Q10 to 2.239 at Q90; DGHE: 1.460 to 1.550). With coefficients maintaining identical signs and monotonic but small changes, how do you justify framing these as “asymmetric” rather than simply demonstrating diminishing marginal effects? The term “asymmetric” typically implies fundamentally different effects across distributions, which your results do not appear to show.
2. Your conceptual framework (Figure 1) positions RQ as a moderator of the expenditure-outcome relationship and a mediator between VA and SDG3. However, your main specification (Equation 1) treats RQ, VA, and GE as simple independent variables with additive effects, completely ignoring the interactive/moderating relationships central to your stated theoretical contribution. Why did you not include interaction terms (e.g., CHE × RQ) in your MMQR specification, and doesn’t this omission undermine your claim that “governance strengthens the impact of health expenditure”?
3. Your policy recommendations suggest different strategies for “lower-ranked countries” versus “higher-ranked countries” based on quantile effects (e.g., p. 33-34 recommending increased DGHE for India). However, the quantiles in your MMQR represent different levels of SDG3 performance, not different countries. Could you clarify how you infer country-specific recommendations from cross-sectional distributional effects when the panel structure means Q10 countries in one year may be Q90 countries in another?
4. Your SEM analysis shows VA has a negative direct effect on SDG3 (-8.03, p<0.001) but a positive indirect effect through RQ (3.32). While you interpret this as partial mediation, standard mediation frameworks typically require a non-significant or positive total effect for meaningful interpretation when direct effects are negative. Given the substantial negative direct effect, isn’t it more accurate to conclude that RQ suppresses rather than mediates the negative VA-SDG3 relationship? This fundamentally changes the policy implications you draw.
5. The manuscript states data “will be made available on request,” but you’ve combined data from the SDG Index, World Development Indicators, and WGI—datasets with different country coverage and reporting standards. Can you confirm that all E7 countries have complete data for all variables for the entire 2000-2023 period, and if not, how did you handle missing observations? This is particularly relevant given the SDG Index only began reporting in 2015.