Earnings Surprise Implicit in Stock Prices: Which Earnings Forecasting Models are Investors Using and What Determines Their Choice?
提出一种新的盈利惊喜度量：隐含盈利惊喜，通过股票价格直接估计随机游走、时间序列和分析师预测的权重，发现机构持股比例低和信息环境差的股票更依赖随机游走预测，且预测准确性影响模型选择，该度量能显著提升基于盈利公告后漂移的投资策略收益。
Abstract: In this paper we address the issue of modeling the relationship between stock prices and accounting earnings in the presence of potential divergence of opinion regarding the expected company earnings. We introduce a new measure of the earnings surprise, the implied earnings surprise, which we define as the weighted average of the random walk, time series and the analysts’ earnings surprises, with the weights being estimated directly from stock prices. The link function which determines the relationship between the stock returns and the implied earnings surprise is estimated semi‐nonparametrically, allowing our framework to nest a variety of models used in previous studies. Our key findings are as follows. First, we find the weight of the random walk (analysts) forecast to be significantly larger (smaller) for the stocks with low share of institutional holdings and impoverished information environment. Second, we find the choice of a particular earnings forecasting model to be related to its forecast accuracy, an effect which is more pronounced for the institutional investors. Finally, we show how conditioning on the implied measure of the earnings surprise substantially improves the profitability of the post‐earnings announcement drift‐based investment strategies