(FORTUNE Magazine) – In our ragged economy the engine that has kept consumers spending and has created more stock market winners than any other industry is residential real estate. The housing boom spread its magic to REITs, homebuilders, and title insurance companies. Nowhere has the bonanza been richer than for providers of the nation's mortgages, from thriving community banks to onetime regionals like Washington Mutual and Wells Fargo that harnessed home loans to reach the top tier in banking. Since 2000 the stocks of the ten financial companies that dominate mortgages have risen an average of 78%, vs. a 30% drop in the S&P 500.

Now fear is rampant that the housing boom is over. The main force propelling residential real estate--the lowest interest rates in 40 years--is suffering a brutal reversal. Since June the rate on 30-year mortgages has gone from just under 5% to 6.1%. A new $500,000 home loan now costs homeowners $2,540 a month in interest, compared with $2,040 in June, an increase of 25%. Suddenly the throngs who rushed to refinance their homes at bargain rates are thinning. Weekly home-loan applications have already shrunk 80% from their peak in May, and that's no blip. The market scaled such outlandish peaks that a nasty fall is all but inevitable. This year the Mortgage Bankers Association forecasts that Americans will pile on $3.2 trillion in new mortgages, beating the record of $2.5 trillion set in 2002; the MBA predicts that mortgage origination will decline 53% to $1.5 trillion in 2004.

So will the financial companies that rode the rocket fizzle along with their best product, mortgages? The best gauge for the impact of the tough new market for home loans is Countrywide Financial (CFC, $67). Countrywide, of Calabasas, Calif., near Los Angeles, is practically a pure play in mortgages. It's the only independent left in an industry now dominated by big, diversified banks, from Wells Fargo to Citigroup. Countrywide covers every part of the mortgage spectrum. It originates home loans, provides servicing, trades mortgage-backed securities, sells title insurance, and even has a bank funded by the $14 billion in property tax and insurance payments that Countrywide holds in escrow for customers.

Countrywide is by far the biggest and most overlooked beneficiary of the home-loan boom. "No bank did a better job mining the great mortgage market than Countrywide," says David Hendler, an analyst at CreditSights. The numbers are extraordinary. This year Countrywide is on track to write $400 billion in home loans, meaning one U.S. household in eight is getting its mortgage at Countrywide. It ranks as the second-largest home-loan provider in America, after Wells Fargo and ahead of Washington Mutual. This year the company and most analysts expect Countrywide to earn $1.9 billion. That approaches what Dell and 3M made last year and far exceeds the profits of Walt Disney and McDonald's.

Most amazing of all is that Countrywide boasts the best stock market performance of any financial services company in the FORTUNE 500, measured from the start of the Great Bull Market over two decades ago. During that period, it has delivered investors a 23,000% return--yes, 23,000%--or 30% a year, so that shareholders now have over $230,000 for every $1,000 they invested in late 1982. It handily beat superstars WaMu, Wal-Mart, and Warren Buffett's Berkshire Hathaway (all around 15,000%). Countrywide's approach is a primer for creating shareholder value. It made virtually no acquisitions and issued relatively tiny amounts of new stock. And it supercharged its shares the old-fashioned way, by producing big gains in earnings with tiny dollops of new capital.

But those numbers bring us back to the great question: Can Countrywide, or any of the major banks that rely heavily on mortgages, possibly perform at even close to the same level in the maelstrom ahead? If Countrywide's market share stays about where it is now, which is quite possible, its earnings will drop 30% over the next couple of years; if it keeps waxing its rivals, its profits will suffer only a slight fall. The possibility of a 30% earnings drop sounds pretty bad, but it's disappointing only when measured against last year's peak. In the worst case, Countrywide's profits will remain far higher than they were in 2002, which was the previous record year.

As for rivals like WaMu, Wells Fargo, Bank of America, and J.P. Morgan, the best bet is that the profits in their mortgage divisions in 2004 will range from a slight decline (Wells and J.P. Morgan) to single-digit increases (WaMu and BofA), a reversal from their big growth rates in 2003. Countrywide got a much larger earnings spike from the new-mortgage explosion than its banking rivals and will suffer a steeper drop in the sluggish market ahead. The reason is simple: Countrywide isn't primarily a bank. It sells most of the mortgages it originates to investors, booking the income immediately, a practice that produces huge profits in great markets. A Wells or Citigroup holds a chunk of its loans on its own balance sheet, generating income over a number of years as the interest payments roll in. Hence, the banks' new mortgage earnings tend to be smoother than Countrywide's.

Despite the rise in rates, the mortgage industry can keep chugging because the housing market's prospects aren't nearly as bad as they look. The rise in rates mainly hits refis, always an erratic, cyclical business. But America's ever-growing population virtually guarantees that builders will keep building new houses, and so the market for mortgages on newly purchased homes should stay strong. That sturdy business consistently grows at 7% to 8% a year, which, accounting for inflation, closely tracks population and GDP growth. In fact, the $1.5 trillion in total new mortgages projected for 2004 would be an excellent number if it weren't such a falloff from 2003: It far exceeds the $1.2-trillion-per-year average of the late 1990s.

Still, that doesn't solve the problem of how to replace the huge jump in earnings from the refi explosion. Countrywide may not replace all those profits, says Stan Kurland, the company's COO, "but servicing can replace much of the earnings lost in originations." The mortgage business has two pillars of support: origination and servicing. Servicing--the mundane business of mailing out mortgage bills, processing checks, and escrowing tax charges--is a powerful hedge against a decline in new loans. Put simply, markets that are bad for new mortgages are good for servicing. As rates rise steeply, as they are doing today, far fewer people prepay their mortgages. That allows Countrywide to collect servicing fees for far longer periods than when falling rates spur homeowners to refinance over and over again. "By strengthening servicing in good times, Countrywide has done a brilliant job of insulating itself for the down cycle," says James A. Johnson, former CEO of Fannie Mae. Countrywide isn't alone in using servicing's countervailing power: Because their profits from new loans will fall less, the big banks can recoup most of the lost earnings from servicing.

Countrywide could even hold its earnings near today's lofty levels if it's able to keep winning market share from big banks. That knack can be traced to CEO Angelo Mozilo, 64, the butcher's son who co-founded Countrywide with $500,000 and three employees 34 years ago. Mozilo has turned the industry's biggest trend--consolidation--to his advantage by not participating in it. He has never bought another mortgage company. "I believe 100% in organic growth," says Mozilo, ensconced in his glass-cube office that juxtaposes sweeping views of the coastal hills with a splendid collection of wintry Hudson River school paintings.

Mozilo has exploited the upheaval caused by a raft of mergers to gain chunks of market share. Countrywide has picked up a slew of star salespeople who didn't like working for a new employer. "Many salespeople were unhappy with the bosses installed by the acquiring companies," says Patrick Burton, an executive recruiter in Portland, Ore. "That's been a big help to Countrywide." Case in point: Countrywide has dined out on WaMu's 2001 acquisition of North American Mortgage, a big player, by hiring no fewer than 300 anxious bankers from the North American network. As a result of its poaching, Countrywide has lifted its market share from 8.2% in early 2002 to 13% today, and it's still gaining.

The bantam-sized, perma-tanned boss is a blend of Jesuit-trained intellectual and unstoppable glad-hander. His internal switch is permanently stuck on "sell"--he sells to bartenders, fellow golfers at tony Sherwood Country Club near L.A., whomever, wherever. Mozilo grew up in the Bronx and started chopping chickens in his dad's shop at age 12. At 14 he took a second after-school job as a messenger at a Manhattan mortgage company. After graduating from Fordham in 1960 with a philosophy degree--"I studied the sages from Aquinas to Schopenhauer, from the most optimistic to the most pessimistic"--he joined the firm full-time. "Between butchering and mortgages," he says, "the choice was clear."

A year later his company merged with a rival in Norfolk, owned by a quantitative genius who would become Mozilo's partner, David Loeb. Loeb dispatched Mozilo to a place he'd never heard of: Orlando. "I thought he said 'Atlanta,'" recalls Mozilo. "The only exciting thing was an occasional rocket going off from Cape Canaveral." Mozilo thrived selling mortgages to homebuilders, surprising the scruffy contractors by appearing at building sites in the same garb he features now, pinstripe suits, white-collared blue shirts, and deer-hunter-orange ties. In 1969 a rival bought the Norfolk company; Mozilo and Loeb quit and founded Countrywide.

Mozilo has always played the role of salesman-in-chief, leaving the number crunching to a crack team of quants now headed by Kurland. (Loeb died in June.) In the very early days Mozilo was literally the sole salesman. When he'd present a bunch of applications to the Veterans Administration, then as now a major funder of mortgages, the uptight officials would routinely turn all the candidates down. "I'd tell the rejected veterans to bring their medals, canes, and artificial limbs and sit in the VA office," says Mozilo. "Then I'd go down and yell at the bureaucrats, 'These people spilled blood for their country! This is a disgrace!'" He adds with an air of triumph: "I'd get almost every application approved."

How well Countrywide fares through the down cycle depends on how much of the earnings decline in new mortgages Mozilo can win back in servicing. When lenders write mortgages, they typically retain the right to service those loans. Homeowners pay about 40 basis points (0.40%) a year--it's folded into the rate on their mortgages--for servicing, or $2,000 a year ($167 a month) on a $500,000 loan. When Countrywide or any other lender takes on servicing for a new mortgage, it must follow GAAP rules by immediately booking an asset and recognizing income, based on the present value of those mortgage servicing rights, or MSRs. To establish the value of the MSRs, Countrywide and its rivals must project how many years a loan will stay on the books.

But as rates fell in 2002 and 2003, Americans rushed to prepay their mortgages. "We reckoned the loans booked in 2000 would last seven years, as in most markets," says Kurland. In fact, they were paid off in two or less. That forced lenders to take huge losses by writing down the value of the MSRs, though profits from new loans more than offset the deficits in servicing.

Now, however, with rising rates, far fewer homeowners are prepaying their loans. Countrywide has grown its servicing portfolio faster than most mortgage companies--from $302 billion in May 2001 to $600 billion today. Until rates stopped falling in June, it had been predicting that those $298 billion in mortgages would get prepaid in two to three years. Now it looks as if they'll stay on the books a lot longer. A mortgage booked in March estimated to last for two years could easily last six. Hence, the cash flows, and profits, from servicing are poised to surge.

Let's go through the numbers. Matt Vetto of Smith Barney argues that the best way to establish a value for Countrywide stock is to forecast how much the company will earn not in an extraordinary year like 2003 but in a normal mortgage market, where originations average about $1.5 trillion. To arrive at the most likely scenario, Vetto created a detailed earnings model using conservative assumptions--specifically, that Countrywide will sustain but no longer grow market share. Mozilo, of course, wants to grow market share dramatically, but the going will be tougher than when the merger frenzy made it easy to pick off top salesmen.

In Vetto's earnings forecast, the revenues from servicing swing from a big loss in 2003 to around $450 million in profits in a normal market. But that isn't enough to compensate for a drop in earnings from origination, from over $2 billion in 2003 to $400 million. With trading, banking, and other sectors kicking in around $430 million, Countrywide should post profits of about $1.3 billion in an average year, vs. about $1.9 billion in 2003, a decline of 30%.

That really isn't so bad. In 2002, a great year for new mortgages, Countrywide earned $842 million. Countrywide now carries a multiple of around seven times its most likely future earnings; big banks like Citigroup and Wells carry P/Es of around 14. So is Countrywide a screaming buy? Probably not. Its profits are bound to be more erratic than those of diversified banks, which have other lines of business to cushion a fall in mortgages. Countrywide's heavy reliance on servicing introduces additional risk: It's extremely difficult to predict how much longer mortgages will stay on the books than the lender originally forecast. On the other hand, the tiny P/E means that a super performance in servicing or continued gains in market share could substantially lift its stock price. So if you're a venturesome investor, trapped in a forest of towering P/Es, you could do worse than pick the unsung maverick of mortgages.

Feedback? investing@fortunemail.com