<p>This study examined the influence of spatial interdependencies on regional total factor productivity in Japan by focusing on the contrasting effects of borrowed size and agglomeration shadows. Using panel data from 47 Japanese prefectures between 2000 and 2018, we compared the manufacturing and non-manufacturing sectors by estimating the Spatial Durbin Model and the Spatial Durbin–SARAR Model. These models enable a detailed decomposition of productivity effects into direct (within region) and indirect (spillover) components across space. The results indicated that industrial agglomeration and transport capital positively influence local productivity in both sectors. However, negative spillovers, interpreted as agglomeration shadow effects, were also observed, where the concentration in neighboring regions suppressed local performance. Human capital exerted a consistently positive impact on manufacturing, with evidence of interregional spillover effects. These findings highlight the sector-specific spatial dynamics and indicate the need for differentiating regional policy responses. For the manufacturing sector, promoting networked industrial clustering and labor mobility can enhance productivity but must be balanced against the risks of overconcentration. In the non-manufacturing sector, decentralized service economies and spatially equitable infrastructure investments are critical for reducing competitive displacement from urban cores. By clarifying the conditions under which borrowed size and agglomeration shadow effects prevail, this study contributes to spatial productivity theory and offers actionable insights for designing regionally differentiated development strategies.</p>