芒格曾经讲过一个故事：“我想从加州的供水系统说起。当初，在设计供水系统时，参考的是很短一段时间内的气象记录。短期的记录很详尽，长期的记录没那么详尽，所以做设计的人只考虑了短期记录。如果把几百年前的情况都考虑在内，他们就会发现，他们设计的供水系统有缺陷，没办法应对严重的旱灾。”其实银行业也一样，罪魁祸首是依赖历史数据进行风险建模，但依赖的历史数据又太短， 去看大萧条时银行是怎么进行资产配置的？去看战争时期银行业的资产负债有什么变化？？去看银行业危机时，全世界那些最优秀的银行是怎么度过危机的，这才能对银行业有一个相对完整的认知。书中举了一个很有意思的例子，席勒与房地美的首席经济学家进行辩论时，后者认为房地美最糟糕的压力测试是房价下跌13.4%，辩称自大萧条以来没有发生过更大的跌幅：
Shiller (2008,p. 52) recalls a debate in 2006 (in the preceding decade house prices had increased by 85%) with the chief economist of Freddie Mac in which the latter defended the fact that Freddie’s worst stress test was for a 13.4% fall in house prices arguing that a bigger drop had not occurred since the Great Depression. I am sure that his statistical analysis is correct but would it not strike you as showing a lack of imagination when the worst situation a big mortgage player can think of is that prices are still up 60% from ten years ago?