Shaopu Wu, Jianguang Hou, Danlin Yu
Sustainable industrial transition requires productivity evidence that separates real efficiency improvement from scale expansion, capital deepening, and reporting change. This study develops a reproducible frontier-productivity diagnostic for provincial leading industry policy, using official 2013–2023 sector panels for 23 two-digit manufacturing sectors in Shandong Province and a matched 2019–2023 benchmark against Jiangsu. The framework combines input-oriented Banker–Charnes–Cooper (BCC) data envelopment analysis (DEA), Simar–Wilson bootstrap bias correction, Malmquist total factor productivity change (TFPCH) decomposition, producer price index (PPI) deflation diagnostics, scale-productivity classification, and targeted sensitivity tests. Bootstrap correction lowers mean BCC efficiency from 0.77 to 0.69 in Shandong and from 0.79 to 0.70 in Jiangsu. Uniform provincial PPI deflation leaves constant-returns-to-scale (CRS) Malmquist estimates almost unchanged, whereas asymmetric deflation creates measurable sensitivity. Direct sector-cluster resampling places Shandong’s aggregate TFPCH at 1.016 with a 95% interval of 0.995–1.045, supporting a near-stationary interpretation rather than a broad upgrading surge; Jiangsu’s corresponding estimate is 0.976 with a 95% interval of 0.955–0.997. The study does not measure environmental performance directly. It shows how frontier-productivity evidence should be stress-tested and paired with environmental indicators before it is used in sustainability-oriented industrial policy.