Tuesday, July 24, 2007

The Once and Future King

Bill Fleckenstein is another of my favorite investment writers, although I suspect that I hardly ever find anything useful in his articles. His weekly jeremiad about the bubble and its impending burst recently included a barely hidden paraphrase of some ideas from Benjamin Graham which are worth amplifying a little, and then trying to apply to a specific case. Mr. Fleckenstein writes:
In the investment business, there are two components of an outcome you expect to see in the marketplace. The first is your analysis of the phenomenon or security you are scrutinizing. The second is your opinion (educated guess) about how other people will greet (price) the outcome that you expect.

In other words, in analyzing a security one starts with what one knows about the current facts, and then continues to what one thinks about future events. One hopes that at least some of the current facts can be objectively known, although in the investing world, even more than in most of our worlds, the 'facts' are often determined by the way in which the questions are asked. It also often happens that what we think are facts are actually what someone else wants us to believe. - Ever hear of financial statements?
Mr. Fleckenstein could have listed a third type of quasi-information, aside from current knowledge and future guesses. He could have mentioned our current decision as to what to do now, based on our guess as to what will happen, which is in turn based on our almost-knowledge of the way things are now.

Now for our example, American Express (AXP). Is it worth buying now, or not?

On July 19, 2007, Jon Markman called American Express an "opportunity of a lifetime". On July 23, Morningstar analyst Michael Kon said that "American Express... shares 'are not a screaming value'". Although Kon's remark was made just after AXP's earnings report, and Markman's was made just before, there's more to the disconnect than that. Look at Markman's reasoning:
Consider this paradox: Investors have given youthful rival MasterCard (MA) valuations ranging from 15 times earnings to 30 times earnings since it went public last year. And just a few weeks ago, the market decided to bestow a valuation of 24 times forward earnings on the long-bumbling provider of the Discover card, Discover Financial Services (DFS), when it split off from Morgan Stanley (MS). Yet Amex -- which has put up years of awesome cash flow and industry-leading return on capital, and which has undimmed future prospects both at home and overseas -- is going for what amounts to just 16 times projected 2008 earnings. This is like pricing a Mercedes-Benz as if it were a Yugo.

In other words, Markman is basing his analysis of the value of AXP mainly on its price relative to its peers. But do you remember the saying, usually attributed to Mr. Graham, that "In the short run, the market is a voting machine but in the long run it is a weighing machine"? And do you remember Mr. Graham's warnings about letting Mr. Market, your counterparty in your securities transactions, price your securities for you? So why is Mr. Markman letting Mr. Market decide what the 'correct' value of AXP is, for the long run? I don't know, but I didn't buy AXP when I saw Mr. Markman's enthusiastic article a week ago, and I doubt that I'll be buying now. I do understand, on the other hand, Mr. Kon's logic: He is using a mathematical model to project an absolute - though very inexact - future value for AXP from its current financial report, and then concludes that his model does not allow enough of a margin of safety to make the stock a bargain today.

Tuesday, July 17, 2007

Parisian Hoodlums and Bloodthirsty Indigenes

As you all know, Jim Jubak of MSN Money is one of my favorite writers on day-to-day investing. But even so, he sometimes uses logic which I find to be a little peculiar, to say the least. Sometimes he uses funny logic so consistently that it's worth discussing in its own right.

One of the weirdnesses which Mr. Jubak performs consistently is to use calculations of a stock's potential value in five, ten, or twenty years as a basis for recommending it for Jubak's Picks, his imaginary portfolio "for a 12- to 18-month time horizon". Recently he used such logic to recommend the oil exploration and production company Apache Corporation (APA). His logic in that article is entirely based on the growing scarcity of cheap oil "over the next five years". Unfortunately, even given the fact that oil exploration stocks usually rise in tandem with oil prices, I don't understand how he knows that oil prices will be high over the next 12-18 months because supplies will be tight over the next five years. The stock market discounts the future, and the oil futures market discounts the future, but neither of them is that efficient or that predictable as a discounting mechanism. Recent experience seems to suggest that significant changes in oil prices over the short term are more likely to have political causes - including causes related to the arcane internal politics of OPEC - than causes related to conventional economics.

Mr. Jubak also considers Apache Corporation undervalued relative to its peers. "The stock is a relative bargain in the oil industry, trading at a forward price-to-earnings ratio of 11.5 versus 13.1 for Devon Energy (DVN) or 12.90 for ExxonMobil (XOM)." If these numbers were very precise, I might say that that's not bad; let's call it, for the sake of argument, a 15% discount relative to its peers. Unfortunately, we're talking about a "forward price to earnings ratio", in other words, an estimate, or a guess. Except for those companies who use 'creative accounting' always to beat the consensus earnings estimates by one cent, it's not clear that the margin of error on these forward estimates is smaller than that discount. My impression is that the anonymous author of an article which recently appeared on Reuters thinks that it is not. In any case, I'd be happier if I could find some evidence that it is not only smaller, but significantly smaller.

There is another repeating problem with using P/E ratios or other standard ratios of value in the context of a 12 to 18 month recommendation: The use of these ratios is based on the idea that earnings somehow positively correlate with stock prices, over the long run, and that therefore as the earnings of a company rise, the price of its stock should also rise, to maintain an 'appropriate' ratio. But what evidence is there that this readjustment takes place within a 12 to 18 month period? Maybe the lag before the famous 'reversion to the mean' is longer. Maybe much longer.

There may be many good reasons to buy Apache, including those listed by Mr. Jubak. But with oil prices having recently reached new local highs for reasons unrelated to Mr. Jubak's long-term calculations, and given that noone will be stunned if stock prices correct somewhat this August or September, perhaps taking Apache's down price with them, I am not well convinced.

to my broker review site