2013 Nobel Prize Revisited: Do Shiller's Models Really Have Predictive Power?. Issue 2 (June 2014)

• Record Type: Journal Article
• Authors: Kantor, Brian; Holdsworth, Christopher
• Is Part Of: Journal of applied corporate finance. Volume 26:Issue 2(2014)
• Page Start: 101
• Page End: 108
• Publication Date: 2014-06
• Subjects: Corporations -- Finance -- Periodicals; Capital investments -- Periodicals; Business planning -- Periodicals; Corporate governance -- Periodicals; 338.6041
• DOI: 10.1111/jacf.12072
• Languages: English
• ISSNs: 1936-8216

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <p>The authors take a critical view of the investment approach advocated by recent Nobel laureate Robert Shiller. A critic of efficient markets theory, Shiller has proposed that investors, when attempting to determine whether the S&P Index is under‐ or overvalued, should use a P/E ratio whose denominator is the 120‐month moving average of the company's EPS. But the authors find that such an approach does not provide consistently superior insights to those provided by conventional PEs—and that, for example, the use of both conventional and Shiller PE multiples would have indicated a highly overvalued S&P not only in early 2000—before the bursting of the dotcom bubble—but also in 1996, when Fed Governor Greenspan spoke prematurely of "irrational exuberance."</p> <p>The authors also show that both the Shiller PE and the conventional PE ratios fail a critical statistical test: they are not mean‐reverting—and as a consequence, both ratios can be expected to indicate either undervaluation or overvaluation for very long periods of time. Complicating matters, current earnings are useful to investors in predicting future stock prices only insofar as they provide a reliable guide to future earnings and cash flows. And as one would expect in competitive capital markets, even perfect foreknowledge of future earnings is not likely to be much help since, according to the authors' analysis, five‐year earnings explain on average less than 20% of the variation in prices over consecutive five‐year periods.</p> </abstract>