Learning about an Infrequent Event: Evidence from Flood Insurance Take-Up in the United States
利用美国大区域洪水和洪水保险保单的面板数据，研究发现洪水后保险购买激增，随后逐渐下降至基线，且非受灾社区购买率仅为受灾社区的三分之一，支持包含遗忘或不完全信息的贝叶斯学习模型。
I examine the learning process that economic agents use to update their expectation of an uncertain and infrequently observed event. I use a new nation-wide panel dataset of large regional floods and flood insurance policies to show that insurance take-up spikes the year after a flood and then steadily declines to baseline. Residents in nonflooded communities in the same television media market increase take-up at one-third the rate of flooded communities. I find that insurance take-up is most consistent with a Bayesian learning model that allows for forgetting or incomplete information about past floods.