[Price-to-Sales ratio] Hi Scott, you kind of struck a nerve by quoting
O'shaughnessy's book "What works on Wall Street." I've read both his books. Although he makes a few good points in each, he contradicts his "tenants" of book 1 in book 2 (and even contradicts some of his book 2 tenants within book 2). Consistency is his main point, but he's not even consistent himself! Book 2 is a brilliant example of how to deceive with statistics. His highly touted P/S ratio < 1.0 screen can get a person into some absolute junk unless they understand how to weed out the junk. I have a real problem with him coming on CNBC stating how his 50-stock portfolio of low-PSR stocks backtested so well without giving categorical caveats and instructions on how to use it.
An individual investor can't buy his 50 stock portfolio (rebalanced annually), and his statistics are absolutely invalid for a typical small portfolio of an individual investor. His low-PSR screen can easily be misapplied, distorted, or incorrectly focused on junk to yield unexpected--and costly--results for the unknowing, free-advice-seeking small investor. O'shaughnessy should not expect that people will buy his book to get more details on exactly what he did, after teasing them with the simplicity and performance of screening by low price-to-sales ratio on a 50 stock portfolio.
Some individual investor listening to O'shaughnessy on CNBC will use some internet pay-per-view stock screening service and randomly pick one or two stocks (from a different database than he used) at the bottom of the Low-Price-to-Sales ratio list, probably selecting a name that sounds "techy", and probably thinking the lower the PSR the better. This is a death wish! Price-to-Sales ratios below 0.3 (bottom decile of 6000 stock universe) have an extremely high correlation with companies having very low or negative operating margins and net margins, very low pretax margins, high debt/equity, and high debt/total capitalization (in other words, I agree with your point about increasing the parameters screened for in a PSR-based screen). These low PSR stocks also have a slightly weaker, but still high, correlation to low ROI, and a significant proportion have negative annual earnings and negative book value growth rate. These are not
stocks people should be buying individually, under the presumption of low risk, which is implied by O'shaughnessy's classification of this strategy as the best single-factor value strategy. The lowest PSR stocks are generally companies with high financial risk (high debt), AND high business risk (inconsistent and unpredictable earnings). Additionally, low Price-to-Sales ratio as a screening criteria has an inherent bias toward picking very small companies. Supplemental screening criteria are definitely necessary in order to reduce, or account for, the bias. PSR can be very misleading if blindly
used as a proxy for value. A company can have a low PSR if they are experiencing cost control problems wherein revenues (sales) may not decline even though earnings, margins, and price are deteriorating (or maybe never were much to speak of).
It's no coincidence that low profit margins, high debt, and low growth rate have strong correlation with companies having low market-price-to-sales ratios. However, as a stand-alone ratio, market price-to-sales ratio (i.e. current market price divided by most recent FY sales per share) is neither a measure nor a gauge for valuation--it is simply a statistic--based on actual price divided by actual sales. 'It is what it is' by coincidence--not because investors drove the price to its current levels to provide a certain perceived fair value for price to sales ratio. To assume that PSR is a stand-alone valuation measure assumes the current market price is at its current level BECAUSE sales are some percent of the price. This would only be true if a majority of the investors gauged value as the ratio of price to sales (which very few do). PE ratios are another story altogether. Regardless of whether a person thinks PE ratio is a true valuation measure, you can't argue with the fact that LOTS of people do use it as such, and it therefore becomes a defacto standard measurement of value--as trivialized as it may be.
MARKET PSR vs. INTRINSIC PSR:
Price-to-sales ratio must be compared to an intrinsic valuation measure in order to signify current valuation levels. Both high and low market-price-to-sales ratio stocks can be overvalued, as well as undervalued, based on fundamentals. Valuation (i.e. acquisition value) is just not as simple and straightforward as a single business ratio.
The PSR can be derived from fundamentals (the "intrinsic PSR"), and this does provide a valuation gauge which, when compared to the market PSR, indicates relative over or under valuation. Using the Gordon growth model, the intrinsic price-to-sales ratio (or the price-to-sales ratio based on fundamentals, if you like) contains profit margin, payout ratio, and growth rate in the numerator, and riskiness (by way of the cost of equity) in the denominator. In other words, Intrinsic Price-to-Sales ratio is low when the combination of Profit margins, Payout ratio, and Growth rate is low, and/or Riskiness is high.
Low market-price-to-sales ratio generally reflects the market's knowledge of these factors indirectly, via a low price. One doesn't need to derive a fundamental (intrinsic) PSR in order to gauge the appropriateness of the market PSR (unless you're buying the whole company). You don't need exact figures for intrinsic PSR to spot a statistical outlier-where low market PSR does signify value (JBIL in early '97 is a good example). By its nature, intrinsic PSR involves some forecasting anyway, so it has a built-in uncertainty. If you
look at the key components of intrinsic PSR, especially profit margins and debt, and contrast those with the market PSR, you come away with a much better feel for the reason behind the low market PSR. A noteworthy find would be a group of stocks with profit margins in the top quartile and market PSR in the second or third decile from the bottom (similar to what you mentioned).
As a side note, the HIGHEST decile of market price-to-sales ratio contains some of the best growth names available (haven't run this screen lately, but 6 months ago it contained Intel, Cisco, Microsoft, Lucent, Ascend, McAfee, Parametric Tech, and Tellabs. And Oh by the way O'Shaughnessy touts Intel and Microsoft as premire companies in his second book! But on the air says he would never own them! Now there's consistency!) As a group, it MAY underperform the lowest decile (depends how big the group is--this makes a BIG difference in the results), but with added selection criteria, there are many, many top-performing companies in this highest-PSR group. I think this speaks volumes about the "coincidence factor" of low market PSR as the best single-factor screen for a 50 stock sample.
dh |
