TY - JOUR T1 - Does Dividend Yield Lose Predictive Ability? <em>Evidence from the Bubbles Period</em> JF - The Journal of Investing SP - 59 LP - 68 DO - 10.3905/joi.2015.24.2.059 VL - 24 IS - 2 AU - Kuang-Fu Cheng AU - Tai-Wei Zhang Y1 - 2015/05/31 UR - https://pm-research.com/content/24/2/59.abstract N2 - This article addresses a puzzle: why dividend yield (DY) has lost its predictive ability since the 1990s. Campbell and Shiller [1988]’s dynamic Gordon model provides a theoretical foundation to explain DY’s predictability of stock returns, however, when the transversality condition fails to hold (that is, when a bubble is present), this implies that DY cannot predict stock returns. Using a recursive test procedure, developed by Phillips et al. [2011], to detect bubbles in the New York Stock Exchange Index, we find stock price bubbles indeed occurred from the end of 1991 and ended in September 2008, the starting date of the financial turmoil triggered by the subprime crisis. Along with major real-world events that influenced financial markets and the early-1990s sharp drop in DY, the empirical evidence coincides with our inference (based on Campbell and Shiller’s model), showing that DY is indeed a useful variable in predicting future stock returns during a no-bubble period, but it loses its predictive ability when bubbles are present.TOPICS: Fundamental equity analysis, financial crises and financial market history ER -