American Express Hiding?

Typically I get a more indepth understanding of a company the longer I study it; this is part of the joy of learning. Not so with American Express (AXP).

Last year, AXP spun off American Express Financial Advisors (AEFA). I supported this move believing the narrowed focus and less capital intensive business model would benefit AXP. But I never imagined that the financial statements would become more opaque.

The first baffling issue is AXP's focus on increasing the return on equity (ROE). In the first paragraph of the 2006 shareholder letter, CEO Kenneth Chenault writes, "In the fourth quarter, we raised our ROE range from 28-30% (to 33-36%), reflecting both our performance since the spin-off and continued confidence in our growth potential."

While ROE provides an excellent way to gauge the capital efficiency of a business, the measure has some real limitations. The most severe limitations of ROEs occur when share issuance or repurchase is high. If large share repurchases occur at a market value in excess of book value, the ROE is raised, almost regardless of business performance. I can point to numerous examples of a poorly performing business with a rising ROE.

Later, in the same letter, the CEO points out that billed business grew by 16%, driven by a 9% increase in cards accompanied by an increase in average spending per card of 7%. From this information, many calculations are straightforward. Yet, my computed numbers increasingly vary from those reported by AXP.

In 2006, AXP reported spending per card to be $11,201 versus my calculation of $8,984. Starting in 2001, my calculation was less than 10% apart, but is now 15%. The same widening has occured in the reported discount rate (2.57%)versus the computed discount rate (2.31%) In 2001, my variance was 8 basis points, but has now moved to 26, resulting in the average discount revenue per card reported at $288 versus a computed $208 per card!

Yet, the worst discrepancy is between reported net fees per card ($35) versus my computed net fees per card ($32). In 2001, my variance was $6 more, but that has now moved to $3 less. This example seems very strange. The net fees on the income statement has not moved significantly for the last six years, and yet the number of basic cards issued has grown strongly. How is it possible that net fees per card have not moved down then?

Despite these issues which have arisen during their transition, I continue to believe that AXP has one of the best franchises in the world and have included a picture of myself happily using My Blue Card in Moorea.

Fairwell to First Data Corporation

In September 2006, First Data Corporation (FDC) spun off Western Union. Then a few weeks ago, FDC announced that it was being bought out by KKR. Now I am moving FDC to the "gone (private), but not forgotten (or public yet - again)" category.

I have studied FDC since the Great Tech Bubble. As we saw the use of more plastic and less cash, a colleague of mine asked, "who makes money on that?" I found out that FDC did - in a big way.

FDC dominates merchant-processing, processing over 50% of the U.S. Visa and Mastercard transactions. FDC's market share in a high fixed cost business gives FDC an enormous competitive advantage. On top of that, FDC is in the best business in Texas (besides Dallas and Houston tollways): helping Mexicans send U.S. earned money back to Mexico. FDC's Western Union dominates this business.

Unfortunately, I never became an owner of FDC stock. My ailment might be considered psychosomatic: I get nosebleeds at p/e ratios over 20. With FDC's average p/e ratio hovering above 25, "bargain" opportunities arose when the p/e dropped all the way to 19 - twice in six years. Nevertheless some investors dove in where I feared treading. One of my cachamim was among them: David Swensen.

Most people had not heard of Mr. Swensen until a recent NYT article featured him. The article told how Yale's endowment had generated an investment return of 16.3% per year under Mr. Swensen's leadership. Such an investment record would send lesser individuals into managing a hedge fund with at least a "2 and 20" fee. Rather than working for Yale for a little over $1 million, he could get paid tens or hundreds of millions of dollars.

But that's not Mr. Swensen. He has explained that his motivation is a "mission" to "make money for financial aid for students." Last year, Yale University's president analyzed their top donors and put Mr. Swensen at the top with a "donation" of $7.8 billion. That sum is the amount of outperformance he has generated relative to his peers at Harvard et al.

His sense of mission translates to what he expects of others. When evaluating a money manager, Steve Cohen, who received a 50% performance fee, he commented "the fees alone are enough to say that I don't want a meeting and there are enough people who put together fair deals." Further, he doesn't go for the "trust me" method of "black box" investing, such as ESL Investments.

The article also describes Mr. Swensen as "not afraid to go where other people don't" which is exactly what I found with his investment in FDC.

Equipped with perfect knowledge of what happened, I went back to my research and reevaluated my calculations. For 2007, Western Union still looks like it's worth $24. The rest of FDC still looks like it's worth $28. Together, they're worth $52. 75% of $52 is $39 - a p/e ratio of 15 - something it never got close to. Maybe it will next time around or maybe, by then, I'll better understand why Mr. Swensen thought he could pay a higher price.