From efficiency measurement to spatial correlation: slacks-based directional distance function and Moran’s I study of Vietnam’s provincial development

Abstract This paper evaluates the efficiency of 63 Vietnamese provinces over the 2018–2023 period by employing a slacks-based directional distance function (SBM-DDF) with bootstrap bias correction and spatial autocorrelation analysis. Using inputs of labor, public expenditure, and capital investment alongside desirable outputs—GRDP, HDI, public revenue and poverty reduction—and explicitly treating undesirable outcomes, the study uncovers three distinct phases of efficiency dynamics: an initial slowdown in 2019, resilience during the COVID-19 period, and a strong post-pandemic recovery by 2023. The average efficiency score rose from 0.843 in 2018 to 0.893 in 2023, with persistent bimodal distributions indicating structural heterogeneity. Spatial analysis confirms significant positive autocorrelation, with Global Moran’s I demonstrating that provincial efficiency is spatially dependent. Local Moran’s I identifies four configurations: High–High clusters in the Southeast and Mekong Delta reflecting agglomeration spillovers; Low–Low clusters in the Central Highlands signaling spatial poverty traps; High–Low efficiency islands achieving superior performance in weak regions; and Low–High provinces failing to capture neighboring spillovers. These findings highlight that infrastructure, institutions and human capital jointly condition efficiency through spatial mechanisms. COVID-19-induced disruptions reveal conditional spatial resilience. Policy implications emphasize differentiated, region-based coordination—strengthening High–High networks, addressing Low–Low constraints, enhancing absorptive capacity in Low–High provinces, and replicating High–Low success models—to foster balanced and inclusive growth.