Abstract
Total agricultural machinery power records installed capacity rather than utilization or engineering efficiency. We examine its association with eight input intensities across 31 provincial-level regions in mainland China. The source panel spans 2000–2024, while the cross-input benchmark uses an exact common sample of 616 province-year observations for 2000–2019. Province and year fixed-effects models control for macroeconomic conditions, crop composition, irrigation, and disaster exposure. Inference combines province-clustered standard errors, Benjamini–Hochberg false-discovery-rate adjustment evaluated at
q
p
= 0.0087;
q
= 0.0695; wild-bootstrap
p
= 0.0005). The phosphate-fertilizer category has a positive point estimate but does not pass the multiplicity and bootstrap checks; total fertilizer, potash fertilizer, compound fertilizer, pesticide, and plastic film are indistinguishable from zero. Pairwise cross-equation tests confirm two unadjusted contrasts involving diesel, but none of the 28 differences survives false-discovery adjustment. Longer windows do not produce a general fertilizer response. The results identify a fuel association and several input-specific patterns; they do not estimate an efficiency-induced rebound or a causal effect of machinery.