I apologize in advance if this post is a little wonkish, but I thought it was an issue important enough to warrant comment. I promise to return to my usually un-wonkish ways in future posts.
Myles Shaver and I wrote a paper several years ago in which we questioned the usefulness of a particular ratio - export intensity (click here for a copy of that paper). Export intensity is simply a ratio of export-to-total sales (expressed formally as export sales divided by total sales, where total sales is domestic sales plus foreign sales). Export intensity, you see, has become one metric by which researchers gauge the export performance of a firm. Our objection to this measure was that when export intensity increases, it can be difficult to determine whether that happens because export sales increase, domestic sales decrease, or some combination of the two (by mathematical construction, both end in the same result). Moreover, because export and domestic sales are interrelated, we must consider their joint impact on each other rather than artificially hold one (e.g., domestic sales) constant.
This is but one example of how ratios make inference difficult. Robert Wiseman (Michigan State University) excellently summed up a broad class of problems with ratios in a working paper entitled “Making Researchers out of Monkeys: On the Misuse of Ratios in Management Research” (contact Prof. Wiseman for a copy). In that paper he details how the use of ratios is not only controversial, but can also lead to statistical estimation and inference problems.
Which brings me (in a long-winded way) to what I want to write about today - ratios used by analysts in finance. Specifically:
1. Price/Earnings - Share price divided by earnings per share
2. Price-to-Book - Share price divided by the book value of the firm
These ratios suffer similar shortcomings to those described above, yet analysts often treat them as if the denominator were fixed.
I’ve had enough of folks in the financial community parroting something to the effect, “Well, its price/earnings is historically low, so that makes this equity a good buy” or “It is trading at less than book value therefore the equity is undervalued.” If I had a dollar for every time I’ve heard something along those lines, I’d be wealthier than Bill Gates.
The real issue is not that the ratios are low, but why the ratios are low.
There are several plausible explanations. One is that the ratio is low because the numerator is low - the equity’s price, for whatever reason, is not in line with what the denominator suggests it ought to be - consistent with what the analysts would like to have you believe, or wish were true. Another is that the ratio is low because the denominator is too high - there are issues with earning estimates and/or the book value of the firm - inconsistent with the views of analysts that bandy this information about. (There are other reasons that these ratios might not be in line with their historical averages - e.g., as a result of structural shifts in the broader economy - but I will not address those here).
Let’s, for example, assume that the latter reason is the “correct” one - that the problem is actually one with the denominator. What then would this suggest? If that’s the case, then it’s not the price of the equity that’s wrong, but rather, estimates of earnings (whether current or forward) that are erroneous. Likewise, it could be that book value is off.
This is not implausible (especially in times such as we are currently living). Don’t forget that many of the same folks who tell us that P/E ratios are low are also those who estimate earnings. Analyst estimates of earnings are notoriously sketchy, either as a result of predictable bias, or simple miscalculation (see Lim, 2001 JOF). Moreover, my hunch (although I can’t provide a cite to any concrete research off-hand) is that analyst estimates of earnings become noisier at economic inflection points.
Book value is not a given either. Book value is simply the best accounting estimate of the value of assets. Who’s to say that the accountants have it right? Let me provide an example - that of banks during the credit crisis. Banks have a book value. It is simply the fair value of the assets that they hold. But we’re really not sure of the “true” value of those assets to the firm unless they have a liquid and transparent market, or until they are sold. So with respect to the stated book value of a bank - does it truly represent the value of those assets, or is it an estimate that could be off? As we’ve learned with many banks over the past year, that value could be off quite significantly. So then if we observe a low price-to-book ratio, it could simply be that the denominator is off - that the assets of the firm are not accurately valued.
Low P/E ratios and low price-to-book ratios are certainly informative, but sometimes not in the way most analysts implicitly assume. It doesn’t necessarily indicate that the equity is undervalued. Rather, price could be the fixed parameter - acting as a signal that earning estimates and/or book values are too high.
Message to analysts: Denominators matter too!
So the next time you hear analysts extol the virtues of a particular equity for the reasons stated above, don’t accept their opinion blindly. Instead of buying their thesis, be sure to seek out the underlying cause for the drop (or rise) in those ratios.