(FORTUNE Magazine) – Back on April 13, 1998, Banc One's CEO, John McCoy, thought he'd earned a page in banking history. That morning McCoy had announced a monumental $30 billion merger with First Chicago, easily outdoing his rival Ed Crutchfield of First Union, whose merger with CoreStates in November 1997 had been only half as big. And with $240 billion in total assets, the new Bank One--spelled conventionally this time--now controlled more money too. You could almost hear the institutional ego swelling out in the heartland.

It didn't swell for long. At more or less the same moment, on the very same floor of New York City's Waldorf-Astoria, Hugh McColl, CEO of NationsBank, stepped in and informed the world of his own historic deal: the purchase of BankAmerica for $60 billion--twice the size of McCoy's merger. Even more impressive, the new entity--now known as Bank of America--would control $570 billion in assets spread from coast to coast, making it the first truly national bank.

That spring was the peak of a wave of banking mergers that had been building since 1985. By 1996, Securities Data reported 70 mergers valued at more than $1 billion; for 1997 and 1998, the total number of mergers was nearly 300. Some of those giants--Citigroup, to name one--have watched their stock soar, but bank stocks overall have risen only 1% since January 1998, something many analysts blame on misbegotten mergers.

The Bank One, First Union, and Bank of America deals sounded so promising that it's hard to imagine they could have gone astray, but the three banks have seen their stock prices collapse by an average of 41% since announcing those deals. All three have missed their original earnings targets (Bank One has missed them disastrously). And all three mergers have created disorder and uncertainty in what had been pretty well-oiled machines. In fact, the only clear financial winners, as we'll see later, were the guys who put the deals together.

"We're coming to the end of a decade of mass consolidation," says Nancy Bush, a bank analyst at Prudential, "and we're asking ourselves, Was there a net gain here? Was the rate of earnings growth enhanced? The answer is no."

So what went wrong? A postmortem of the blunders and conflicts at Bank One, First Union, and Bank of America goes a long way toward unraveling the answer--not only for those banks but also for many of their overgrown, underperforming peers. In the end, the lessons are surprisingly simple--elementary, even. Which makes the missteps of McCoy, Crutchfield, and McColl all the more astonishing.

BIG ISN'T SO BEAUTIFUL. Back in the early 1990s many of the deals launched by Banc One, First Union, and NationsBank (now Bank of America) looked nothing short of brilliant--especially as their asset size skyrocketed. Both banking deregulation and the savings-and-loan crisis meant that struggling regional banks could be picked up dirt cheap. So that's essentially what each of the three banks did. By the mid-1990s all had been transformed into multi-state powerhouses.

That's where the trouble began. Bank executives were coming to believe that size was the answer to everything. Larger banks, they figured, could take advantage of economies of scale and make a mint on fees by cross-selling an extensive arsenal of products. "There were people who believed that bigger was always better," says Dick Kovacevich, CEO of Wells Fargo. "You had to reach some mammoth size--over $200 billion or $300 billion--in order to keep up with things like technology. That's a bunch of bunk."

By the end of 1999 all three banks had fattened up like the Nutty Professor. Their assets had grown at an annual average clip of 24% since 1993--more than three times the average rate of all FDIC-insured banks. But that growth certainly didn't extend to their stock: Their stock prices consistently underperformed the BKX bank index's return of 191% over six years, with Bank One's stock rising by only 5%, First Union's by 49%, and Bank of America's by 83%. As Tom Brown of Bankstocks.com puts it, "Basically, when banks get bigger, it's just the shareholder who gets screwed."

AT WHAT PRICE ACQUISITION? Indeed, shareholders didn't seem like much of a priority at any of the three banks, which were too busy trying to win the next big deal--and paying whatever it took to close it. In August 1996, NationsBank shocked Wall Street by laying out a total of $9.7 billion for Boatmen's Bancshares in St. Louis, then the second-highest price ever paid for a U.S. bank. At 2.6 times the book value, Boatmen's wasn't terribly pricey (banks typically pay two to three times book value for their acquisitions), but analysts like Tom Brown still considered it "pretty expensive for a plain-vanilla franchise in a below-value market." Five months later analysts were clucking their tongues again when Banc One announced a $7.3 billion deal with First USA. While the total tab wasn't quite as high as the Boatmen's deal, the premium was outrageous: 43% above First USA's existing stock price, 5.7 times its book value.

Then, in August 1997, NationsBank slammed through the biggest U.S. deal ever. It bought Florida's Barnett Banks for four times book value; the price ran to $15.5 billion. But as Hugh McColl said in discussing whether prices had gotten too high, "Let me ask the question another way: What is the price of not making an acquisition?"

Perhaps that's what Ed Crutchfield was thinking when First Union purchased CoreStates in November 1997 for a record-breaking $17 billion, 5.3 times book. And for what? CoreStates' quarterly earnings had remained flat for the past year. But by the time Crutchfield came along the bidding had become so frenzied that it just didn't matter. As Kovacevich puts it, "At that time we weren't just losing out on deals by 5%, we were losing by 20% to 25% every time. When that happens, somebody has to be wrong."

PROMISES, PROMISES. After spending so obscenely, none of the banks could afford to be wrong. So they primed the pump on Wall Street by promising that the new deals would generate spectacular earnings growth. When Banc One bought First USA back in 1997, it projected that combined earnings would rise at a lovely 16.6% clip through 2000--comfortably higher than its traditional projected growth rate. To make that happen, First USA would simply have to grow by 23%--virtually forever. "In order to justify the premium they paid, Banc One had to give a big earnings target," says Duke Buchan of Maverick Capital. "So McCoy said, 'We need 20% from First USA.' Of course, as the business got bigger, that 20% became a lot tougher to compound."

That's for sure. Rather than growing spectacularly, First USA wound up floundering spectacularly in 1999--brought down at least in part by its own sleazy marketing tactics. In a desperate attempt to woo customers, the credit card company had offered a teaser rate of 3.9%, only to jack it up to 9.9% a few months later. And anyone missing a payment twice in a six-month period--even by just a day--saw his rate spike to 19%. Nice. No wonder First USA's attrition rate nearly doubled, to 16%, between 1998 and 1999. And no wonder Banc One's after-tax operating earnings were only $3.46 per share, 26% below its original target, set in 1997. So much for the First USA growth engine.

First Union planned to ramp up earnings growth as well--only it was going to do so by squeezing every penny from its acquisitions. First Union planned to cut CoreStates' annual expenses by a stringent 45% even though CoreStates was already quite efficient. "If you say you're going to cut costs by 50%, you're telling people one of two things," says Sean Ryan of Byrne Ryan & Co., a research firm. "Either you're saying, 'Our bank is inept and inefficient but should be bigger,' or it's a lie in the first place."

Ineptitude was certainly a factor. To reduce costs as promised, First Union had to close branches and cut back on employees. At the same time it launched a Buck Rogers-like program called Future Bank--a battery of ATMs, telephones, and personal computers. In theory, Future Bank was to free up the remaining people to peddle First Union's new products. But the reality was embarrassing: First Union lost 19% of CoreStates' original account base between April and November 1998. And since then it has hired back some employees to bolster service. "I've heard numerous complaints about how crummy First Union's service was after the merger," says Jim Schmidt, who runs the John Hancock Bank & Thrift fund. "You create a lot of animosity when going through restructuring; most of the time top management doesn't see that."

At first glance Bank of America might seem to be in better shape than Bank One or First Union. After all, its latest round of after-tax operating earnings seem positively rosy--jumping by 29%, to $4.68 per share, from the previous year. But don't be too impressed. Those numbers have undergone some serious cosmetic surgery. For instance, Bank of America's loan-loss provision for 1999 came down by 38%, to $1.8 billion--mainly since last year's number was described as disproportionately high because of "global uncertainty" in 1998. In the fourth quarter, the bank also picked up an after-tax gain of around $80 million on sales of credit-card and other consumer-finance assets. "Overall," complains Credit Suisse First Boston analyst Michael Mayo, "the quality of these earnings is weak." And Bank of America still fell 16% short of the $5.59-per-share pro forma earnings it offered at the time of the NationsBank-BankAmerica deal.

Nor does the future look much better: Revenues picked up by only 6% in 1999, and rising interest rates could spell trouble for Bank of America this year, given that more than half its revenue comes from interest-based income. (Banks like Bank of New York and Chase Manhattan generate around 60% of their revenues from stable, fee-based activity.) "It's a slow leak at Bank of America," says ING Barings' Andrew Collins.

ACCOUNTING TO WALL STREET. As ever, fancy accounting has also been invaluable to the banks in giving their pricey acquisitions a low profile. All three typically relied on what's known as "pooling of interest" accounting. Used just for stock transactions, pooling of interest allows the acquirer to add to its books only the book value of the acquired company--not the full price it paid for the target. That's a pretty neat trick if you've paid a fat premium for an acquisition, as Banc One did for First USA (43% over the market value), First Union did for CoreStates (18% over), and NationsBank did for Barnett (37% over). In the magic of pooling-of-interest accounting, those premiums simply vanish.

What's so useful about that? Well, because the acquisition appears to add little to the surviving company's equity base--even as it captures all the extra earnings from the acquired company--the new bank's return on equity looks as if it's on steroids. The effect is anything but trivial. For example, when Michael Mayo of Credit Suisse First Boston recalculated each bank's cash return on average tangible equity as if it had used purchase accounting, Bank One's 1998 return on equity went from 27% to 12%. At First Union, it fell from 35% to 11.8%. And Bank of America's 29% return on equity dropped to about 10.8%. As Dale Wettlaufer of Legg Mason says, "The end result is that the cash return on tangible equity is a totally bankrupt measure."

That's not to say that Wall Street has been fooled. "For the most part, everyone sees through those return on equity numbers, and they know these banks aren't doing well," says Sean Ryan of Byrne Ryan. And the Financial Accounting Standards Board (FASB) has gotten so tired of inflated deals like these that it's reviewing pooling accounting and may eliminate it by 2001. No wonder all three banks have such abysmal stock performance.

WHO'S REALLY THE BOSS? If there's a key to making sense of these various factors--particularly the banks' hell-bent acquisitiveness on the one hand and their eagerness to downplay the price of that acquisitiveness on the other--it lies in a certain, well, tension between the financial interests of the banks' shareholders and those of their CEOs. As it happens, CEOs are paid partly as a function of asset size--and mergers by definition make that number bigger. A study by Indiana University's Richard Rosen and Babson College's Richard Bliss shows that from 1986 to 1995, mergers increased the median bank CEO's total compensation by an average of 20% to 25% in the three years following a major merger. "And even though stock prices usually fall in the years following big mergers," says Rosen, "few of the CEOs have experienced a significant decline in overall wealth."

The experience of our fearless CEOs certainly echoes this pattern. While their awards or bonuses have come down since, each CEO saw substantial economic gain in the year he pulled off a large merger. For instance, First Union's stock has suffered since it merged with CoreStates, but the deal made a killing for Ed Crutchfield: nearly $20 million in total compensation for 1997, considerably more than he had earned in the years before or has since. John McCoy, now unemployed, snagged a 13% pay hike (excluding options) the year of the First Chicago deal. And the Barnett deal helped boost Hugh McColl's bonus by 13%, to $3.5 million, in 1997 (capital markets problems scaled back his 1998 earnings).

Meanwhile, bank boards are notoriously weak and incestuously close to their CEOs. "That's one of the reasons that bank stocks look so lousy," complains Bush of Prudential. Historically, all banks were cozy, local entities, and CEOs stacked their boards accordingly; take a look at Bank of America's and First Union's board lists, and you'll notice that many of the members hail from the banks' home state of North Carolina. "Ed Crutchfield is close to many of his board members," says Jim Schmidt of John Hancock. "I don't think he's too worried about his job." Analysts also claim that Hugh McColl always made sure he controlled his board: Out of the 19 members of the board last year, 11 were on the old NationsBank board in 1997.

John McCoy's case is somewhat different. Banc One's original board wasn't much different from McColl's or Crutchfield's--it was loaded with several of his old acquaintances from the Columbus, Ohio, country-club circuit. However, when Banc One merged with First Chicago, part of the deal was that the board would be split evenly. Apparently the two sides have been fighting bitterly ever since--and John McCoy has now retired. "The only way the board could settle its dispute was for McCoy to leave," explains Bush. Rumor has it that Citigroup wunderkind Jamie Dimon is being considered as a replacement. Others wonder whether Citigroup itself might not just buy all of Bank One, though one insider says the departure of McCoy--who is reportedly close to Citi's John Reed--may jeopardize that plan.

Of course, bank mergers haven't always spelled disaster, and Citi is a prime example. Citigroup is larger than any of the three we've been discussing, but when Travelers merged with Citicorp in April 1998, handing over $70 billion in stock, no premium was paid: "Since neither side took a lot of dilution, no one had to dream up reasons about how they would win it back," explains Dave Berry, an analyst for Keefe Bruyette. And then Citi wound up saving a reported $2 billion in costs anyway. Its stock is up 65% since 1998.

Meanwhile, Norwest, based in Minneapolis, calmly sat out the most rabid part of the bidding. Only when it became clear that California's Wells Fargo had botched its deal with First Interstate did Norwest swoop in and buy it, for close to market value. Not bad shopping. After that, CEO Kovacevich set about building the banks' customer service; people are now expecting Wells Fargo to pick up some serious California market share--creating yet another problem for Bank of America. Wells Fargo stock has improved by 11% over the past year.

If a more vigorous board at Bank One is one encouraging sign of (incremental) change, at least a couple of other signs have begun cropping up elsewhere. For example, in addition to FASB's possible initiation of accounting reforms, California pension giant Calpers recently listed First Union as a potential activist target, which could jar other large institutional investors into action.

If anything's clear in all this, it's that shareholders have every right to be unhappy with the performance they've gotten out of these supposed juggernauts. In fact, it's a wonder they're still shareholders at all.