Cromwell's Rule and the Role of the Prior in the Economic Metric
展示贝叶斯决策理论中先验密度的设定如何影响经济显著性推断，以投资组合问题为例，说明忽略零假设上的概率点质量会高估预测模型的经济意义，并解释统计与经济显著性不一致的原因。
We show that using Bayesian decision theory to draw inference about the economic significance of a model requires careful specification of model uncertainty in the prior density. As an example, we use the Bayesian investor's portfolio allocation problem to show that failure to include probability point mass on the null hypothesis that returns are not predictable will overstate the economic significance of the predictive model. The dissonance between the statistical and economic significance of asset return predictability seen in previous research appears to be an artifact of how model uncertainty is treated in the specification of the prior.