The Risk in the Magic Formula
October 11th, 2010Joel Greenblatt’s Magic Formula, popularized in The Little Book that Beats the Market, is a simple(ish) two-factor stock screening method that, according to backtest, outperforms other screening methods, such as low P/E.
I like the Formula, and use it. In my opinion, though, it does have one big risk worth correcting for: the Magic Formula rewards a weak balance sheet.
As an example, imagine two widget makers, Silverback Inc. and Consolidated Schmendricks. The two are nearly identical — same products, sales, income statement, balance sheet – with one difference: Schmendricks is currently stiffing its suppliers, so its accounts payable are double those of Silverback’s.
The Magic Formula judges business quality by return on employed capital:
ROEC = EBIT/(current assets – extra cash – current liabilities + PP&E)
In the example, Schmendricks will have much higher return on capital, because it is defaulting on accounts payable. Thus it will be ranked much higher by the Magic Formula.
I’ve chosen an extreme example to illustrate the principle, but the problem exists even away from the extremes. For example, suppose Schmendricks were funding its operations entirely with commercial paper instead of cash. Same problem: the pile of short-term liabilities result in a weaker balance sheet, but a higher Magic Formula rank. In general, the Magic Formula does not consider credit risk.
One might reasonably argue, “but if this is such a problem, why didn’t it show up in the backtests? Why didn’t higher default rates reduce the historical performance of the Magic Formula?” The answer is that there is something missing from the backtest data set: over the tested period, the cost of capital was falling almost monotonically, and there was no credit crisis.
Under those happy conditions, on average, the more levered company wins: they can always roll debt over, and usually at a lower price, so insolvency risk remains low, almost without regard for capital structure.
Unfortunately for the Formula and for all investors, the cost of capital bottomed out at almost the same time the Little Book was released. Suddenly, credit risk is a consideration again.
As a result, I suspect the Magic Formula will not perform as well when the cost of capital is volatile and generally rising, unless one adds an additional factor to control for credit quality. Since the cost of capital is now much more uncertain, and looks likely to rise almost monotonically in the next 30 years, one might want to consider credit rating in choosing among the stocks that rank well under the Magic Formula.
Note that simply screening for low P/E has the same problem. Credit risk generally did not matter for 30 years, but now it does. Value investors in the 1970s did not use P/E alone, but rather a combination of P/E and financial strength. We are headed there again, I suspect.
[...] And speaking of the Magic Formula, here’s a good recent post by Greg Speicher, EV/EBIT Ratio Trumps P/E Ratio, and here is Bill Mitchell from Gemfinder on The Risk in the Magic Formula. [...]
I agree. Curiously though, in the just published update The Little Book That Still Beats The Market Greenblatt’s formula for ROC is a little different to the one you quote. Specifically he excludes short-term debt from current liabilities, which partly gets around the problem of credit risk (i.e. the instance you quote of a company funding itself using commercial paper). I’d be very interested to know if the original edition (which I no longer have) deducts all current liabilities from current assets, or just payables, like the new edition.
[...] had plenty, over £13m, on its balance sheet date. Sharelockholmes uses the formula quoted in this critique, but the new edition of Greenblatt’s The Little Book That Beats the Market, The Little Book that [...]
Bill,
First, I’d like to commend you on the posts so far. I enjoyed reading what you have so far, seeing as it is difficult to find true quality in most places on the internet.
This article caught my interest in particular. Mr. Greenblatt’s formula was something I found quite interesting last year, and although I would never use it myself, it is quite fun to see stocks that I’ve discovered also end up in his screener.
The cost of debt was something I hadn’t thought about when looking at his backtested results, as well as bad balance sheets. The first time I went through his screener, I eliminated any companies with bad balance sheets because I am naturally risk-averse and do not trust highly leveraged companies.
It will be interesting to see how his formula works the next 20 years! thanks for the post.
Feel free to contact me or check out my blog. The items of most value to you might be in my “Articles” section. I have written up about the importance of working capital, operating leverage, my valuation technique, as well as the use of a Du Pont Model. These are four topics I found myself trying to explain to others over and over again, so if you don’t have a good background with any of those topics, it might be interesting for you to read. They are somewhat elementary in nature, but can be used with force if applied correctly.
Do you have articles from awhile back? Or is this a recently-launched website?
Andrew
Bill,
Great quality in your posts, I really enjoy what have to say- will keep an eye on your website for the future. This article was my particular favorite, definitely not something I had considered when I first saw Greenblatt’s magic formula… I’m operate a very concentrated portfolio, so although his screener is fun to check as part of my search strategy for new companies, there would be no chance of me investing in a highly leveraged company anyway. I’m sure there are many novice investors who have signed up for his program and you might want to consider speaking up.
Have you contacted him about your concerns? Do you know if he has received criticism like this already? I figure he would probably like to hear from you about this.
Great comments. Sorry for the delayed response — everything was caught in an overly aggressive spam filter.
Andrew, good suggestion. I’ll contact Joel Greenblatt directly to invite comment.
Richard, very interesting point about the altered Magic Formula in the second edition of the Little Book. My post is quoting from the original edition, in which the formula is laid out in the appendix, pp. 138-144. It would be interesting to know whether the formula itself was altered between editions, or whether it is just being explained in greater detail in the second version.
Yes, as you say, deducting short-term debt will in some cases control for the problem I’m pointing out. A company that is about to fail a debt rollover, for example.
But in other cases it won’t save you. In the example I gave, Consolidated Schmendrick’s is defaulting on accounts payable. This is a trade obligation, a different class from debt.
A simple alternative would be to choose a minimum corporate credit rating that you are willing to accept. Then just use the credit rating as a post-filter to your results from the Magic Formula. It would be interesting to test how this would have performed during the financial crisis.
This reminds me of an amusing tale from a guy I know who once had lunch with Warren Buffett. I’ll relate it in a future post.
While this article is true, the fact is that very few Magic Formula screened stocks have poor balance sheets. I’ve looked at literally hundreds over the past several years and the number of truly weak balance sheets is probably below 10.
I believe for all practical purposes this is a theoretical risk, not an actual one.
That’s a legitimate criticism: this was an illustration without example or proof.
More conclusive would be to backtest the Magic Formula during an extended period of generally rising cost of capital. My intuition is that, under such circumstances, the Formula would do well, but the Formula plus a credit rating factor would do better.
The hard part is how to backtest it. We would need a complete fundamental data set from an extended period of rising capital costs, such as the 1970s. That set would be expensive, if indeed it exists. Any volunteers?
Would it be appropriate to substitute PP&E for Goodwill/Intangibles in certain industries that rely heavily on patents/IP? For example – biotechs or media companies.