American Express's Counter-Cyclical Balance Sheet Is Underappreciated

Summary

• Write-offs will increase, but on a lower base on loans and receivables.
• Lower loans and receivables release capital, allowing American Express to survive a worst-case scenario with adequate capital.
• The company currently trades for about 9x my estimate of normalized pre-tax earnings.

American Express (NYSE:AXP) was founded 170 years ago and still pervades our daily lives. That puts it in uncommon company. Its premium clientele, low credit costs, and high discount rates are well-known and properly appreciated by investors. But one attribute is underappreciated: American Express's balance sheet is counter-cyclical. Although the company's net write-off rate will rise this year, write-offs will be on lower loans and receivables. As loans and receivables decrease, American Express generates capital, fortifying its position.

Cyclical companies are best valued on their through-cycle earnings. In this article, I lay out three simple ways to estimate American Express's normalized, through-cycle earnings power. I explore the implications of the company's counter-cyclical balance sheet, and finally, discuss whether American Express can withstand a worst-case scenario.

My conclusion is that American Express can, and will, survive this storm and is trading at approximately 9x pre-tax earnings or 11.5x after-tax. The business looks poised to return 11% annually over the long term. The stock could outperform the business by 4% annually if it re-rates to 14x, its long-term median, over 5 years. This means it could double by 2025.

Cyclical But Consistent

Although American Express is cyclical, it worth pointing out that it rarely loses money. Profits oscillate between merely good and excellent.

Since 2005, when American Express spun off Ameriprise Financial, it has averaged a 27% ROE. That's an incredible result considering the period spanned the worst financial crisis in a generation. The company's lowest ROE was 15% in 2009. There are not too many businesses in the world whose "worst" year is a 15% return.

(Source: Author, Data from annual reports)

Berkshire Hathaway's 1987 annual report cites a Fortune study which shows just how remarkable this record is. The study examined the 1,000 largest American companies between 1977 and 1986. It found that only 25 of the 1,000 averaged over a 20% ROE and had no single year lower than a 15% ROE. Further, it found that 24 of these 25 "business superstars" outperformed the S&P 500 during that decade. This bodes well for American Express.

Counter-Cyclical Balance Sheet

During times of stress, businesses and consumers lean on their banks to extend more credit. We saw this in Q1: loans grew 8% at Bank of America (BAC) and 6% at JPMorgan (JPM). Loans extended during recessions tend to experience disproportionate write-offs, so this rapid loan growth isn't as desirable as it sounds.

However, American Express's loans and receivables actually decrease in a recession as consumers and businesses pare back spending on their own. The bank's loans are, on balance, profitable, so this reduces earnings. But it also reduces risk at precisely the right time.

(Source: Author, Data from annual reports)

Loans dropped 6%, peak to trough, between 2007 and 2009. They roared back to new highs in 2010. The blip in 2015 and 2016 was the result of selling the Costco portfolio.

Loans in Q1 were down 3% year over year and 10% quarter over quarter. Receivables in Q1 were down 21% year over year and 22% quarter over quarter. These declines are more severe than 2008 by a wide margin. Lockdowns have caused customers to pull back spending much quicker than in 2008.

This has two important implications:

1. Although net write-off rates will rise this fall, they will be on a smaller base. If loans and receivables continue to fall at a double-digit pace, American Express's existing reserves will cover a larger portion of loans than they appear to today.
2. Decreasing loan balances release capital and fortify American Express's financial strength.

Normalized Earnings Power

Despite its consistently high returns, American Express is a cyclical business that should be valued on through-cycle earnings.

I use three methods to triangulate normalized earning power. These methods are simplistic and intended to be directionally correct, if not precise.

The first method is to normalize earnings based on net write-offs. Write-offs averaged 3.3% of loans and 2.4% of receivables since 2005. Applied to $83 billion of loans and $45 billion receivables, normalized credit costs should be $4.0 billion. Subtracting this from 2019's pre-tax pre-provision earnings gets us $8.0 billion of normalized pre-tax income.

The second method is to normalize earnings based on provisions. Provisions averaged 2.8% of loans and receivables since 2005. Applied to $128 billion of loans and receivables, normalized credit costs should be $3.6 billion. Normalized pre-tax earnings would be $8.4 billion.

These two methods apply average credit costs to current loans and receivables. This should prove conservative, since loans and receivables will continue to decrease if the economy remains stalled.

The third method is to normalize earnings based on ROE. Since 2005, American Express has averaged a 27% ROE. Before the financial crisis, its ROE was in the thirties, but low interest rates have kept in it the twenties since. ROE averaged 25% since 2008, which seems indicative of the immediate future. Applying this to $21 billion of equity and backing out taxes at 21% yields $6.6 of normalized pre-tax income.

These estimates place normalized earnings power in the range of $6.6 billion to $8.4 billion, and average $7.7 billion. These are through-cycle estimates, so they're lower than 2019's earnings and probably higher than 2020's.

Estimating Forward Returns

Historically, American Express has retained 20% of net income and reinvested it at a 25% incremental rate. This has allowed it to grow 5% annually. It returns the remaining 80% of earnings to shareholders: 20% as dividends and 60% as buybacks.

At $86 per share, the company is worth $70 billion. This is 9x normalized pre-tax income of $7.7 billion and 11.5x normalized after-tax income.

Buybacks tend to be pro-cyclical at American Express, so it is best to assume they're done at the company's median P/E of 14x. That implies a 4% buyback yield. Added to a 2% dividend yield, investors can expect about 6% back annually from the company.

Altogether we get 5% growth, 4% buyback yield, and 2% dividend yield. This sums to an 11% return before any change in valuation multiples.

If the stock re-rates from 11.5x to 14.0x over 5 years, returns will increase 4% per year, to 15%. That means American Express stock could double from here over the next five years.

What's The Worst-Case Scenario?

The worst-case scenario for American Express is probably that travel and entertainment (T&E) spending goes to zero, while net write-off rates hit their 2009 peak (8.5% of loans, 3.8% of receivables).

Management said that T&E volumes account for 30% of spending on the Q1 call. So far, they're down 95%. If revenue decreases 30% in 2020, it would come in at $30 billion.

Some expenses, like cardmember rewards, cardmember services, and business development, naturally decrease when cardmember spending declines. Management thinks that these three will decrease with a 50% beta to revenue (Q1 call). To be conservative, let's assume that all other expenses remain at their 2019 levels. Management has committed to zero layoffs in 2020, so it's a reasonable assumption for salaries. Management plans to stop marketing and customer acquisition activities, so I'll assume those expenses are down 75% (zero from April through year-end). These assumptions get us to $4 billion of pre-tax pre-provision earnings.

(Source: Author, 2019 annual report, Q1 2020 call)

If credit costs equal their worst in 2009, without further building reserves, they'd come in at $9 billion.

(Source: Author, 2020 Q1 10-Q)

Altogether, American Express would report about a $4.6 billion loss. This would be notably worse than anything the company experienced during the financial crisis. It wouldn't be good, but it would not be a disaster either.

How would this affect American Express's capital? The company targets a 10-11% CET1 ratio and ended Q1 at 11.7%. It must maintain a 6.5% CET1 ratio to be considered "well-capitalized." In the worst-case scenario I've outlined, its CET1 capital would decrease by approximately $4.6 billion.

RWA is harder to predict. I assumed a 30% decline in loans and receivables, and I'd expect RWA to track this nearly one-for-one. To be ultraconservative, however, let's assume RWA declines at only half that rate. In that scenario, American Express's CET1 ratio would fall from 11.7% to 10.1%. That's still within management's internal target and well above the regulatory minimum.

Incidentally, this is about where the company's capital ratios bottomed out during the financial crisis. Its CET1 ratio was 9.8% at the end of 2009 and 11.1% by the end of 2010. This is American Express's counter-cyclical balance sheet in action.

Risks

A quick recovery will benefit American Express. The longer lockdowns linger and travel suffers, the more the company will hurt. From my view, American Express will almost surely survive the crisis. So, the risk is not bankruptcy but impaired earnings power.

I assumed earnings normalize at a 25% ROE. But if the economy stagnates, ROE could linger at 2009's 15%. That would imply $4 billion of pre-tax income. At 14x earnings, that equals a $55 stock price. American Express traded into the high $60s in March, so seeing this price isn't totally out of the realm of possibility.

Conclusion

American Express is a remarkable company with a differentiated business model and a long history of success. This is not its first stress test, though this one is putting unique demands on the company.

A worst-case scenario suggests that the company will have adequate capital to survive. Though earnings will be below average the next year or two, long-term investors should focus on the company's through-cycle earnings power.

Today, the stock trades for 9x normalized pre-tax earnings power. Buffett has made a career over buying high-return businesses for 9-10x pre-tax income.

The business looks poised to compound at about 11% and the stock could compound at 15%, doubling over the next five years. The quicker the recovery, the better the company will do.

Analyst’s Disclosure: I am/we are long AXP. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Comments

[Librarian Capital]

I like the approach of trying to estimate AXP's normalised earnings. However:

Of the 3 methods used, the first two start with a similar loans & receivables number as AXP has today. This seems to assume the non-interest part of the P&L stays the same - which implicitly assumes the 30% of the billed business that is T&E just goes back to normal. It also ignores the hit to AXP's capital base that the article went on to discuss in method 3.

Method 3 starts with an estimates of AXP's post-downturn capital base, which the article believes will first suffer a $4.6bn in the worst case. (I don't think this is the worst case.) $4.6bn is a large number compared to the latest CET1 capital figure of $17.3bn. The article may be right that credit card loans being paid off would automatically raise the CET1 ratio and remove the need of new fundraising - but AXP will need to raise capital if they want their loan balance and thus earnings to return to their former shape (as methods 1, 2 imply).

The actual calculations in method 3 basically assume $21bn of equity (the same as current) and a 25% ROE. The $21bn assumes no hit to AXP's capital base, which I think is doubtful; the 25% ROE figure, while it seems superficially conservative compared to the 2019 figure of 30%, ignores how AXP's ROE has been trending down fast (it was 33.5% for 2018) due to its capital-intensive growth with loans. (The worst case also starts with a $21bn equity, ignoring the $4.6bn loss the article itself discussed, so it doesn't seem really the worst case.)

The big question for AXP is when things normalise - billed business was actually down 45% year-on-year in April, because in addition to T&E down 95% non-T&E was down significantly too. And, with 30% of billed business from T&E (including 7% from airlines), and the business model being quite leveraged, there is a good chance that even a 15% ROE may prove too optimistic, let alone a 25% one.

[Matt Franz]

Thanks for your detailed comment. You're right - there are two variables here. Credit costs are rising while loans are falling. If you look out 5-10 years, I don't think you need to worry too much about what the exact mix will be next year. There is a wide distribution of outcomes for next year, which if why I took a stab at a worst case scenario. I think my worst case scenario is reasonable in that RWA will probably contract faster than I assumed, which leaves room for greater credit costs than assumed.

My expectation is for a U shaped recovery. If billed business recovers in two years, as it did in '09 and '15 then the stock should do well. If we're in the early days of a depression, it won't.

In a U shaped recovery AXP will need to retain capital to fund new loans. Buybacks will most likely remain on hold. If this happens, discount revenue will return and generate a lot of capital. They might need 1-2 years to rebuild capital.

Overall, I think the distribution of short-terms is too wide to handicap. If you think AXP can survive long enough to get to the long-term, then their long-term earnings power is substantial. The risk seems to be dead money for a few years while they rebuild.

[Alfonso Benz]

Great article. I would like to add that American Express achieves a 10% yield on its loans. So, if it reached a delinquency of 8% as in 2008, it would still be less than the earnings from the loans. the data is $ 80kM in loans and $ 8kM in gross interest earnings. Regarding the evaluation, if we look at the operating cash flow the company is trading at only x5. I recently bought. I share a post on the subject.
inbestia.com/...

[MasterLucas]

right mindset bro i bought in the high 60s but it is now to cheap too