Two banks, one start line: JPMorgan, Citigroup and the arithmetic of survival
Run the clock back to 31 December 1999. You have $20,000 to divide between the two most consequential names in American banking. Citigroup looks like the clear choice. It is the world’s largest financial firm, earning $13.5 billion a year at a 22% return on equity. The bank about to take the JPMorgan name earns under half as much, at a lower return, with a market value around one-third of Citi’s. You put $10,000 into each and leave them alone for twenty-six years.
By the end of 2025, with dividends reinvested, the JPMorgan holding is worth about $131,000. The Citigroup holding is worth about $5,300, original stake included. The S&P 500 turned $10,000 into roughly $75,000. The stronger bank of 1999 lost 47% of the money, while its weaker-looking peer compounded at 10.4% a year. Both faced the same industry, regulators, crises and starting date.
View the data
| Year-end | JPMorgan | Citigroup | S&P 500 |
|---|---|---|---|
| 1999 | $10,000 | $10,000 | $10,000 |
| 2004 | $9,100 | $13,700 | $9,100 |
| 2008 | $8,400 | $2,300 | $7,300 |
| 2012 | $12,600 | $1,300 | $12,600 |
| 2016 | $27,500 | $2,100 | $21,400 |
| 2020 | $45,300 | $2,400 | $38,500 |
| 2024 | $95,400 | $3,100 | $63,900 |
| 2025 | $130,900 | $5,300 | $75,300 |
Sandy Weill built Citi as the “financial supermarket”, combining banking, insurance and broking in one group, and business schools studied it. Its management has spent two decades taking it apart. Travelers left in 2002, Smith Barney was sold in the crisis, and Jane Fraser has withdrawn consumer banking from fourteen countries. Fraser has said Citi is “no longer the financial supermarket of the past” (Source).
How losses become permanent
The crisis hit every bank. JPMorgan shares were down 57% on a total-return basis from the 1999 start to the March 2009 trough. What separated the two was what each had to do to survive, and what that did to the original owner’s claim.
Citigroup lost $27.7 billion in 2008 and needed $45 billion of TARP capital plus a government guarantee over $306 billion of troubled assets. In 2009 the government converted preferred shares into common at $3.25 and ended up owning roughly 27% of the company. Basic shares outstanding rose from around 5.3 billion to roughly 28.8 billion by 2010. When earnings recovered, they were split across five times as many shares. The dividend, the equivalent of $21.60 a year in 2007 on today’s share basis, is now $2.68.
JPMorgan entered 2007 with capital near 7%, when many competitors ran at 3–4%. Jamie Dimon made this discipline famous as the “fortress balance sheet”. It cut its dividend in 2009 and took TARP money it quickly repaid, but it never made the government a co-owner or multiplied its share count. Its spare capacity let it buy while others sold: Bear Stearns in 2008, Washington Mutual’s banking operations that September, First Republic in 2023. The dividend is now $6.60 a year.
Return on equity shows the same split. Citigroup earned the higher return every year from 2000 to 2006, while the supermarket was still being praised. JPMorgan has earned the higher return every year since 2007.
View the data
| Year | JPMorgan | Citigroup |
|---|---|---|
| 2000 | 15.56% | 22.4% |
| 2001 | 3.92% | 19.7% |
| 2002 | 4% | 18.6% |
| 2003 | 16% | 19.8% |
| 2004 | 6% | 17% |
| 2005 | 8% | 22.3% |
| 2006 | 13% | 18.8% |
| 2007 | 13% | 2.9% |
| 2008 | 4% | -28.8% |
| 2009 | 6% | -9.4% |
| 2010 | 10% | 6.8% |
| 2011 | 11% | 6.3% |
| 2012 | 11% | 4.1% |
| 2013 | 9% | 7% |
| 2014 | 10% | 3.4% |
| 2015 | 11% | 8.1% |
| 2016 | 10% | 6.6% |
| 2017 | 10% | -3.9% |
| 2018 | 13% | 9.4% |
| 2019 | 15% | 10.3% |
| 2020 | 12% | 5.7% |
| 2021 | 19% | 11.5% |
| 2022 | 14% | 7.7% |
| 2023 | 17% | 4.3% |
| 2024 | 18% | 6.1% |
| 2025 | 17% | 6.8% |
Preparation and succession
In July 2007 Citi’s chief executive, Chuck Prince, told the Financial Times that “as long as the music is playing, you’ve got to get up and dance” (Source). Four months later Citi disclosed up to $11 billion of further subprime write-downs and Prince resigned. JPMorgan had begun pulling back from subprime securitisation in late 2006, giving up fees while competitors still collected them.
JPMorgan has had one chief executive since the end of 2005. Citi has had five: Prince, Bischoff, Pandit, Corbat and Fraser. Weill had earlier pushed out credible successors, among them a younger Jamie Dimon, whom Citi fired in 1998. Dimon rebuilt Bank One and arrived at JPMorgan through the 2004 merger.
Why the gap persists
Fraser has delivered one of banking’s better turnarounds. Citi returned $17.6 billion to shareholders in 2025, and its shares returned 42% in 2024 and 70% in 2025. Yet the 1999 shareholder was still down 47% at the end of 2025. Including dividends, Citi’s total return fell 98% from peak to trough, and a 98% loss needs a 4,900% gain to get back to the start. JPMorgan suffered drawdowns of 74% and 68%, but never recapitalised at the bottom, so every recovery belonged to the same owners.
The week of 13 October
The third-quarter results both banks published on 13 October 2026 capture the history in miniature. Citi made $5.2 billion, up 40% on a year earlier, with its best third-quarter revenue in a decade. Return on tangible common equity was 11.5%, and Citi returned $4.1 billion to common shareholders, 85% of their earnings (Source).
JPMorgan made $15.3 billion, about three times as much, at a 20% return on tangible common equity. It also cut net buybacks to $664 million: “In line with our hierarchy, we reduced share repurchases this quarter” (Source). Dimon listed elevated asset prices among the risks to watch.
Tangible book value per share sums up the twenty-six years: $112.64 at JPMorgan and $101.59 at Citi. The capital per share is almost the same. JPMorgan earns around 20% on it, Citi about 11.5%, and JPMorgan is the one holding back while asset prices look high.
The lessons for public companies
Strip out the banking and the lessons fall into three groups: balance sheet, capital allocation and leadership.
The balance sheet
-
Survival is the first return. Citi owners did not lose a quarter century because the bank earned poorly on average. They lost it in about eighteen months, once. The discipline that prevents this is built in calm and held under pressure. It looks inefficient, lowers returns and hands an advantage to bolder competitors in every year except the decisive one.
-
Dilution is how a bad year becomes a bad quarter century. Losses in the income statement can heal. Equity issued at the bottom stays issued, and the share count divides every future dividend, buyback and earnings figure.
Capital allocation
-
The round trip in your own shares is the cardinal sin. Citi bought back stock above $45 in the boom and issued billions of shares at $3.25 in the bust, the reverse of the Buffett test, which says to buy below intrinsic value and issue above it. JPMorgan used the same tool in the opposite order, and its third-quarter cut shows it still treats buybacks as the last claim on capital.
-
Transact from strength, or be transacted upon. JPMorgan built its modern franchise by buying in storms at prices only a fortress could pay. Citi built the supermarket at the top of the cycle and sold Smith Barney, a crown jewel, in the crisis to raise capital. Strong companies choose when to deal. Weak ones take the timing they are given.
-
A dividend is a residual, not a promise. Citi kept its boom-level dividend until January 2008, months after the write-downs began, then cut it in forced stages to one penny. JPMorgan cut early, while still profitable, called it insurance and restored the payment within two years. The difference was who made the cut, management or events.
Leadership
-
Succession is capital allocation. One company kept one chief executive for two decades. The other went through five, having first fired the executive who went on to build its rival. Boards often treat succession as an HR process. On this evidence it ranks among a company’s most consequential capital decisions, because only a secure chief executive can defend an unfashionable capital position through a full boom.
-
The objective function is the strategy. Weill set out to build the largest financial firm in the world, and did. Dimon’s shareholder letters keep returning to value per share, with the fortress as the means. Both teams delivered what they aimed at. A board should ask what a chief executive is maximising before it admires the execution.
-
Candour is a risk system. Prince’s “still dancing” described a culture that could not stop while competitors kept earning. JPMorgan’s retreat from subprime in late 2006 meant forgoing fees and looking wrong for the last stretch of a boom. Leaders need room to do that.
-
Complexity compounds too, as a liability. The supermarket failed slowly and then expensively. The synergies never came, management could not see into parts of the business, and regulators eventually found Citi unable to govern its own data. If a successor’s obvious first move is to dismantle what you built, shareholders will eventually pay for the dismantling.
The investor who picked Citigroup in 1999 was not foolish. They bought higher profit, a stronger return on equity, a celebrated strategy and the consensus of the day. Capital kept in the good years, complexity tolerated for too long and executive succession opened the gap. The crisis both banks shared only exposed those decisions.
The scoreboard that counts is the one a patient owner reads twenty-six years later, and it keeps a perfect memory of the year management would most like to forget.
Methodology: total returns run from 31 December 1999 to 31 December 2025, dividends reinvested, in nominal US dollars, from dividend- and split-adjusted prices (Yahoo Finance), cross-checked against independent return calculations. Citi’s figures allow for the 2002 Travelers spin-off and the 2011 reverse split. Before December 2000 the JPMorgan series is Chase Manhattan. Fundamentals come from the banks’ filings.