Technological Innovation, Frontier Advancement, and Green Development Efficiency: Evidence from Mountainous Counties in China

Ecologically fragile and economically underdeveloped mountainous regions face the dual challenge of pursuing catch-up development while maintaining ecological security. Although scientific and technological innovation is widely viewed as a key driver of green transition, county-level evidence remains limited on whether, how, and under what conditions innovation improves green development efficiency (GDE) in such constrained settings. Using a balanced panel of 62 mountainous counties in Hebei Province, China, from 2007 to 2022, this study measures GDE with a super-efficiency SBM model, constructs a composite index of technological innovation with the entropy weight method, and estimates dynamic panel models to examine the direct effect, mediating pathway, and heterogeneity of technological innovation. The results are as follows: (1) Technological innovation significantly promotes GDE, but the effect is mainly released through a lagged and cumulative process rather than fully in the contemporaneous period. This finding suggests that innovation dividends in mountainous counties depend on local technology absorption, organizational adjustment, and complementary infrastructure. (2) Mechanism identification confirms that frontier advancement capability, captured by the Technical Change (TC) index derived from the Malmquist–Luenberger framework, serves as a crucial endogenous mediation channel. Innovation systematically lifts the green efficiency ceiling by accelerating the outward shift in the production possibility frontier toward pollution and carbon reduction, yielding a highly significant indirect effect. (3) The innovation–GDE relationship is conditioned by institutional settings and local absorptive capacity. The effect is strongest in counties included in the 2010 cohort of National Key Ecological Function Zones, whereas in low-efficiency counties the innovation effect is positive but weaker and less persistent, so short-term gains do not automatically accumulate into long-term efficiency advantages. These findings provide county-level empirical evidence for designing time-phased and institutionally differentiated green innovation policies in ecologically constrained, underdeveloped regions.