Stock Returns and Dividend Yields Revisited: A New Way to Look at an Old Problem
提出一种名为子抽样的新统计方法，用于在弱条件下构建置信区间，并应用于股息收益率能否预测股票回报的问题，基于三个数据集未发现令人信服的预测证据。
We i n troduce a new statistical method for nding good con dence intervals for unknown parameters in the context of dependent and possibly heteroskedastic random variables, called subsampling.It works under very weak conditions and avoids the pitfalls of having to choose a structural model to t to observed data.Appropriate simulation studies suggest that is has better small sample properties than the GMM method, which a l s o w orks under weak conditions and is model-free.We use the subsampling method to discuss the problem of whether stock returns can be predicted from dividend yields.Looking at three data sets, we do not nd convincing evidence for predictability o f s t o c k returns.