Answers are grounded in this paper and cite their sources.
Introduction Understanding why green innovation diffuses unevenly across firm populations requires a model that separates the structure of firms’ knowledge from the networks through which they learn. Methods We develop an agent-based model in which heterogeneous firms search a dual-dimensional NK landscape comprising conventional and green knowledge, acquire external knowledge through a learning network constructed relative to a competition-reference topology, and respond to carbon pricing, green R&D subsidies, and targeted learning-edge additions. The cross-domain parameter C specifies a one-way dependence of green fitness contributions on conventional knowledge. We distinguish firm-level adoption, system transition, and persistent lock-in, report uncertainty using 100 replications for ordinary conditions and 500 for focal lock-in estimates, and distinguish implementation verification, conceptual and structural validation, literature-grounded qualitative pattern comparison, and the limits of predictive validation. Results In 100-replication Monte Carlo experiments, final mean green fitness under a carbon tax falls from 0.713 at C = 0 to 0.614 at C = 5 , while the final firm-level adoption rate, defined by a green-fitness threshold, falls from 0.941 to 0.762. Homogeneous-behaviour and mechanism-ablation benchmarks show that behavioural heterogeneity materially changes aggregate and cross-firm outcomes, internal search accounts for most of the tax response, and imitation can also propagate comparatively weak knowledge configurations. Carbon pricing is the most consistently effective single instrument. Targeted bridge addition has little effect without an incentive change and yields only a small, knowledge-structure-dependent incremental effect when combined with taxation. Discussion The diagnostics are consistent with contextual revaluation of green configurations and changes in the conversion of search proposals into implemented improvements, without separating their contributions to the aggregate decline. Because the model is stylised rather than industry calibrated, its policy results are interpreted as conditional mechanism evidence rather than sector-level predictions.
No takes yet. Share an insight, caveat, or question.
郝闻达 et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: