Multiples for Valuation: Go High, Go Low, Ignore the Middle

Multiples such as the dividend yield (D/P), the price-earnings ratio (P/E), and the cyclically adjusted price-earnings ratio (CAPE) have long been viewed as being useful to forecast returns over periods of 10 or so years. The evidence discussed in this article supports this belief and takes it one step further by showing that multiples are far more useful when they are relatively high or low than when they are somewhere in the middle of their historical range. In fact, relatively high or low multiples are more highly correlated to forward returns in sample, and produce better return forecasts out of sample, than multiples that lie somewhere in the middle.