While a number of empirical studies have explored the determining factors of welfare caseloads, none of them has examined the effect of workload on caseload. However, several studies outside the field of economics have suggested that workload may be an important factor in determining caseload size, in that higher workloads may lead to the rationing of assistance. This would mean that a greater number of caseloads per caseworker should decrease total caseloads. Using a panel of Japanese cities, this paper estimates the effect of workload on caseload size to examine whether the rationing of social assistance benefits does occur. The results support for the existence of the rationing. This study also examines the effects of caseload size on the number of caseworkers to see how localities adjust their caseworkers to increasing needs of social assistance. The estimation finds that the adjustment is quite sluggish. On average, the localities may well not even employ one additional caseworker, even if their caseload increases by almost 100.