Warren Buffett knows a thing or two about banks. His portfolio has long been stocked with them and if it wasn’t for a 1970 amendment to the Bank Holding Company Act, he would likely have owned them outright, too, as he does in insurance. 

“We feel it’s something…that falls within our circle of competence to evaluate… We don’t think it’s beyond us to understand the banking business,” he told attendees at his annual general meeting in 1995. 

So when he buys a new banking stock, it’s worth paying attention. This week, regulatory disclosures revealed that Buffett bought $1 billion worth of Capital One in the first quarter. To make room, he sold out of another banking stock – US Bancorp, a top ten holding for ten years until the middle of last year. 

Buffett is no stranger to credit card businesses like Capital One. American Express is one of his longest held names. But unlike Amex, Capital One has morphed into a full-fledged bank and it has done so relatively quickly. When Buffett opened his current position in Amex in 1991, Capital One had not yet even been formed. 

To try and understand what Buffett sees in Capital One, let’s take a quick look at where the company came from, and where it stands today.

Warren Buffett is the longest serving chairman and CEO of a financial company in America. Second is Richard Fairbank, chairman and CEO of Capital One.

In the mid-1980s, Fairbank was a consultant to the financial services industry. Working out of the Watergate Building in Washington, DC, he and colleague Nigel Morris would delve into bank balance sheets to see where clients were making their money. Consistently, one business line stood out: credit cards. “It was growing at 20 and 30 percent a year, and it was making 30 to 40 percent on equity,” Nigel Morris recounted.

The credit card business hadn’t changed much since its emergence thirty years earlier. Card issuers generally had one or two card products (e.g., a classic card and/or a gold card) each of which charged a uniform annual percentage rate of around 18%. If an applicant could pass the risk threshold set by the issuer, they would receive a card; if their credit behaviour was deemed too risky, their application was rejected. This resulted in a portfolio of customers priced as if they had very similar probabilities of default.

Fairbank thought the business could be run better. He identified a steep gradient of customer profitability underneath the average return and spied an opportunity to boost both growth and profitability by introducing more granular pricing. As the son of two successful physicists, he saw potential to inject scientific rigour into the business. “The industry lends itself to massive scientific testing because it has millions of customers and a very flexible product where the terms and nature of the product can be individualised,” he told students at Stanford University.

He and Morris toured the country, meeting over twenty heads of credit card divisions and CEOs to sell them on the idea. None bought it. Profitability in the category was already good enough for most and they didn’t want to disturb it. One CEO even threatened to “throw Fairbank out of the window if he ever again recommended a business with charge-off rates over one percent.” 

Eventually, one bank warmed to the idea: Signet Bank of Richmond, Virginia (now part of Wells Fargo). The only catch: its management wanted Fairbank and Morris to come over and run the project. The pair agreed and in October 1989 joined Signet to refashion its credit card business.

When the first credit card was launched in Fresno, California in 1958, it was a disaster. The bank behind it – Bank of America – had predicted that delinquencies would average 4%. Yet after issuing around 2 million cards, delinquencies rocketed to 22%. Fraud was rampant and no provision for collections had been made. Its foray into credit cards cost Bank of America $20 million including advertising and overhead – no small amount in the late 1950s.

Fairbank and Morris had a similarly torrid time when they tested their new credit card model. Signet’s charge-off rate rose from 2%, among the industry’s best, to more than 6%, among the industry’s worst. 

But, like Bank of America, they stuck with it, and performance quickly improved. The breakthrough came via a test which involved transferring a cardholder’s other credit card balances to Signet, for which the customer would receive a low introductory rate. In 1992, the bank rolled out this teaser offer nationwide and the response was outstanding. At one point, Signet had to hire 100 people a week just to handle balance transfer applications.

By 1994, credit cards made up two-thirds of Signet’s group revenue, up from a quarter in 1988, and were the driver of significant share price appreciation: from the time the new card offer was introduced through to October 1994, Signet was the best performing stock on the New York Stock Exchange. In order to release more value, Signet agreed to spin-off the card business into a new company, Capital One, with Fairbank at the helm.

As an independent entity, Capital One had to grapple with the challenge of marketing its product. Competitors had clear advantages that underpinned their market position. American Express had its brand; Citibank economies of scale; and MBNA links to affinity groups. Capital One had its “information-based strategy” but that wasn’t sufficient to win new customers. 

So, from the very beginning, Capital One spent heavily on marketing. In its first two years as a stand-alone company, it doubled marketing spend to over $200 million (a figure which today stands at over $4 billion). Some of that was channelled to TV and billboard advertising, but a lot was spent on direct marketing. In 2002, the group cut a deal with the US Postal Service for bulk mailing and sent an unprecedented number of direct mailings out, offering a mix of competitive teaser rates, lower annual percentage rates and more options.

Meanwhile, testing continued in an effort to innovate new products and win new customers. In 1999, the group performed 36,000 tests. By then its marketing budget had grown to $750 million, of which 80% was spent on marketing products that didn’t exist a year before. In its first five years post-IPO, Capital One generated almost $850 million in profits, giving it plenty of firepower to invest in growth.

Eventually, Capital One would come to dominate the credit card market. 

Today, Capital One is the third largest card issuer in the US by outstanding balances, with a share in excess of 10%. While early growth stemmed from targeting customers who were being overcharged by other issuers relative to their risk profile, later growth came from subprime segments as Capital One welcomed borrowers otherwise denied access to credit. That legacy is still reflected in the group’s business mix, with around a third of its card receivables coming from subprime borrowers (compared with 15% at Chase and Citi, the two largest players). 

In its (US) card business, the group earns a revenue yield of around 18% – not far off the 18% average rate the industry pulled in 35 years ago, albeit with more dispersion and a different revenue mix underpinning it. Unlike players with a more payments-oriented focus, Capital One earns the bulk of its income via interest payments. Including international operations, the group booked just $4.6 billion in interchange fees last year, versus $16.6 billion of net interest income from cards. 

Marketing can be flexed up and down but total operating expenses consume around half of US card revenues, leaving a profit before credit loss equivalent to around 9% of loan balances. It’s a number that’s been pretty stable, sitting in a range between 8.1% and 9.1% over the past ten years.

Beneath that line is where the action happens. Credit charge-offs take a big, but variable, bite out of operating income. In the last downturn in 2009, losses peaked at 9.2%. They troughed at 1.90% in 2021, but are on the rise again. Latest monthly data for April has losses up at 4.3% and early warning 30 day delinquencies are back to pre-Covid levels. The big question facing Capital One today is how its card portfolio will perform in a slowing economic environment.

The key driver of charge-offs is the unemployment rate. Fairbank calls unemployment and card charge-offs “two humps of the camel”. In the past, he’s pointed out a one-to-one relationship between them. “Unemployment goes up a couple 100 basis points, industry credit card charge-offs have sort of done the same,” he observed in 2009. This time, he’s more focused on the rate of change. 

Jeff Norris from the company’s finance team explains: “One intuitive way to think about that is it’s kind of the formation of new unemployment that is the bigger driver of card charge-offs. When you think about if somebody loses their job, it will take them six months to roll through the delinquency buckets and charge-off at 180 days.”

Right now, the company projects that US unemployment will rise to 5% by the end of 2023 or the beginning of 2024, from 3.4% currently. That’s a pretty dramatic increase. The average forecast among economists is 4.3% for the end of 2023, rising to 4.7% by the middle of 2024. Many economists, including those at Goldman Sachs, are forecasting unemployment not to rise very much at all. In anticipation of its bearish outlook, Capital One has raised a lot of reserves, adding over $2 billion to its allowance for US card losses since the beginning of 2022. 

If unemployment doesn’t rise that quickly, Capital One may be over-reserved.

One thing the company doesn’t have to worry about in the current environment is funding. When it launched, Capital One was heavily reliant on capital markets to fund its credit card receivables. But after entering into a memorandum of understanding with bank regulators over certain regulatory matters in 2002, the markets got spooked. 

The group had opened a small bank several years earlier, but deposits stood at only $17.3 billion against balance sheet loans of $27.9 billion (plus a further $31.9 billion securitised loans). Management was able to avert a liquidity crisis by cutting back on asset growth and aggressively marketing web-based deposits as an alternative funding source.

The episode led to a rethink. “In good times,” Fairbank said, “the capital markets are an efficient funding source. But in bad times, we believed an economic downturn could create funding challenges during an extended period of capital markets dislocation. Funding is like oxygen. Having it 98% of the time just isn’t good enough.”

Fairbank anyway wanted to diversify into other areas of financial services. He had experimented with mobile phones and vacation clubs earlier, but closed them down in favour of finance. In 1998, he acquired a Dallas-based auto finance company and in 2001 bought an online provider of direct motor vehicle loans. In 2005, he went full-steam into banking, acquiring New Orleans-based Hibernia National Bank for $5 billion and later New York’s North Fork Bancorp for $15 billion. After the financial crisis, he also bought ING Direct USA for $9 billion. 

The deposits these banks brought with them underpin the group’s funding. Unlike at other banks, deposits grew in the first quarter, to $350 billion, exceeding outstanding loans which stand at $295 billion. The bulk of these deposits are interest-bearing, so, at 2.2%, their average cost is higher than at some other banks but compared with alternative funding sources (see More Net Interest below) they remain good value. 

By contrast, some of the loans these banks brought turned out not to be so good. A $1 billion taxi medallion portfolio inherited from North Fork turned sour when Uber came onto the scene. A $3.7 billion oil and gas portfolio inherited from Hibernia got hit when borrowers in that industry got squeezed in 2015. And now the group’s $3.6 billion exposure to commercial office property is under pressure. 

Looking through these portfolios, though, the overall loan book is currently performing fine. And try as it might to diversify away from them, credit cards are tough to dilute. Although domestic cards now make up only 42% of Capital One’s loan portfolio, they generate 56% of the earnings (last 12 months). Throw in international cards and other consumer lending like auto and the share of the loan book grows to 70% while the share of earnings grows to 90%. 

Capital One has attracted many copycats over the years. Tinkoff in Russia and Nubank in Brazil were both modelled on the company, their founders realising that credit cards can provide a platform for profitable growth. As a shareholder of Nubank, Warren Buffett will know this. Spotting a disconnect in valuation, he’s gone back to the source...