Value Investing's Long Run

What the gradual turn away from modern portfolio theory holds for value investing.

The finance discipline is in the process of a halting transition. The efficient markets/modern portfolio theory is giving way to broader perspectives that incorporate the realities of information asymmetry — the fact that all market participants do not have the same access to relevant information — and deeply ingrained behavioral biases that often dominate actual financial market outcomes. At the leading business schools these latter approaches are now firmly established even though in the finance profession at large they remain relatively unfamiliar.

One particular aspect of this change is the increasing importance of value investing, an approach to investment management pioneered by Benjamin Graham and David Dodd ’21 at Columbia. In part, this is due to the overwhelming success of the value approach in practice. Individual value investors like Warren Buffett and value-oriented institutions like Sanford Bernstein populate the ranks of outstanding investors out of all proportion to their numbers. However, it is also due to a detailed appreciation of the way value investing differs from more conventional approaches, how this is responsible for the historical success of value investors, and why it is likely to continue in the future.

Traditional characterizations of value investing have been overly simplistic. Value investing involves buying securities at one-third or greater discounts to their “true” values, effectively buying dollar bills for fifty cents. More recently we have begun to appreciate how a value approach is distinct in the particular areas of searching for investment opportunities and valuing companies.

In search, the value strategy is to look in areas that are obscure, boring, unattractive, and therefore cheap by common metrics like low market-to-book and PE ratios. Simple statistically constructed portfolios of such stocks produce above average returns in cross-sections of securities over all extended time periods in all global markets. (My colleague Tano Santos shows this in his take on the question.) In time-series, there are reliably predictive levels of positive serial correlation in short-term returns. Neither phenomenon should be observed if markets were perfectly efficient. (Attempts have been made to associate the higher cross-sectional returns of “cheap” stocks with higher levels of risk, but these risk factors never turn out to be empirically measurable independent of cheapness.) Behavioral research phenomena such as loss-aversion, overconfidence, and lottery preference — the disproportionate desirability of low probability, very high reward outcomes — account for the value of “cheapness” much more directly. These behavioral approaches suggest that value strategies that focus on undesirable, boring, ugly, and hence, “cheap” areas like distressed securities deserve special attentions, and are likely to continue to be successful as long as the underlying behaviors persist.

In valuation, the discounted cash flow (DCF) approach that business school teach their students has three obvious shortcomings. First, it generally fails to make use of balance sheet information and an approach that ignores potentially significant information will be inferior to an approach that does not. Second, a DCF calculation is a weighted sum of future cash flow estimates. It involves adding good information — the value estimates of near-term cash flows — to bad information — the value estimates of far-future cash flows (typically embodied in a terminal value). The result, as any engineer knows, is that the bad information dominates. A DCF calculation never segregates estimated elements of value by degree of reliability so that reliable elements of value can be separated from unreliable ones. Third, the input assumptions to DCF valuations are parametric — profit margins, revenue levels, growth rates, capital intensities, and costs of capital. There is no easy way to integrate strategic judgments — whether an industry will be viable in the future or whether any firms are likely to enjoy sustainable competitive advantages — into a DCF calculation.

The value approach to valuation — starting with the most reliable information, the balance sheet, to obtain an asset value, then looking at the value of a firm’s present day earnings power, and only then looking at the value of future growth — suffers from none of these deficiencies. It uses all the information (including the balance sheet), organizes that information from most to least reliable (balance sheet to current earnings to future growth in earnings), and is based on a clear relationship between strategic industry judgments and valuation. For example, if a firm does not enjoy competitive advantages in its markets, then it will never sustainably earn above its cost of capital on investments in growth. Under these circumstances growth creates no value and the growth element of value can be ignored no matter how high the growth rate is.

Notwithstanding the statistical evidence, there is an inescapable sense in which markets are efficient. The average return of all investors (before fees) must equal the average return on all assets. Any gains relative to the average by one investor must be offset by the losses of another. Better approaches to search and valuation have placed — and will continue to place — most value investors firmly in the above market part of the return distribution.

About the researcher

Bruce Greenwald

Professor Bruce C. N. Greenwald holds the Robert Heilbrunn Professorship of Finance and Asset Management at Columbia Business School and is the academic Director...

Read more.