Flood Shocks, Public Service Capacity, and Agricultural Total Factor Productivity Change: Evidence from China and ASEAN

Climate-induced floods increasingly disrupt agricultural production systems, posing escalating threats to food security and economic stability in climate-exposed and trade-integrated regions. This study examines the relationship between flood shocks and agricultural total factor productivity (TFP) change across China and ten ASEAN economies and investigates whether public service capacity and structural conditions attenuate this negative association. A country–year panel dataset covering 1993–2024 is constructed by aggregating event-level flood records from EM-DAT and matching them with World Bank indicators of agricultural production, population, public services, and industrial structure. Agricultural TFP change is measured using the DEA-Malmquist index, with arable land, agricultural land, agricultural employment, fertilizer use, agricultural freshwater use, and rural electricity access as inputs and crop production and agricultural value added as outputs. A two-way fixed-effects model is then applied to examine the conditional association between flood mortality severity and agricultural TFP change. The empirical results show that greater flood mortality severity is significantly associated with lower agricultural TFP change. This negative contemporaneous association remains evident across the reported alternative specifications and sensitivity analyses. Current health expenditure per capita, energy use per capita, access to basic drinking-water services, and medium- and high-tech manufacturing value added are positively associated with an attenuation of the negative association between flood mortality severity and agricultural TFP change. These findings suggest that agricultural productivity performance under climate-related disturbances is shaped not only by exposure to flood shocks but also by policy-relevant capacities embedded in public service provision and structural conditions. A notable finding is that the marginal association becomes positive and statistically significant at high observed levels of health expenditure per capita and energy use per capita, whereas the corresponding estimates for drinking-water services and medium- and high-technology manufacturing remain statistically indistinguishable from zero.