Minimal dataset—How digital proximity shapes firms’ optimal energy demand: Evidence from Chinese manufacturing

With respect to sample selection, we retain only manufacturing firms and exclude those with severe missing values in energy consumption data. Firms flagged as *ST, ST, or PT, which indicate financial distress or delisting risk, are removed from the sample. In addition, continuous variables are winsorized at the 1% and 99% levels to mitigate the influence of extreme values. After applying these procedures, the final dataset comprises an unbalanced panel of 2,942 Chinese manufacturing firms covering the period 2014 to 2024.