Does digital information use improve technical efficiency? evidence from potato smallholders in China

Abstract This study aims to analyze the impact of digital information use on technical efficiency (TE) in potato production, using a two-year panel dataset comprising 1,332 observations from smallholders in China. To address potential selection bias, a propensity score matching method combined with a selectivity-corrected stochastic frontier analysis is employed. A random-effects panel Tobit model and an instrumental variable (IV) mediation model are further used to explore the direct and indirect effects of digital information use on TE. After correcting selectivity bias, the average TE equals 0.529, with significantly higher efficiency observed among smallholders who use digital information. The results of the Tobit model indicate that digital information use significantly enhances the TE of potato production, with a coefficient of 0.060. The IV mediation analysis reveals that this effect operates through three key behavioral channels: risk reduction, input optimization, and market participation. Furthermore, heterogeneity analysis shows that the positive effect of digital information use on TE is stronger among smallholders with limited social capital or distant market access. Policy implications for enhancing smallholders’ use to digital information through targeted support are discussed, with the aim of improving technical efficiency.