This study re-examines the buffer stock hypothesis regarding livestock by taking into account differences in wealth level, asset types, and periods after a shock. This paper takes advantage of a unique panel data set of agricultural households in Southern Province, Zambia. The data were collected by weekly interviews of 48 sample households from November 2007 to December 2009, covering two crop years in which an unusually heavy rainfall event took place. If we consider delayed responses to the heavy rain shock, our econometric analyses support the buffer stock hypothesis for cattle as well as small livestock. Overall, this paper suggests that conventional annual data sets used in the existing literature may miss the period-dependent transactions of assets after a shock.