The Beautiful Machine

=20

The = Beautiful=20 Machine
Greed on Wall Street and blindness in Washington = certainly=20 helped cause the financial system's crash. But a deeper explanation = begins 20=20 years ago with a bold experiment to master the variable that has = defeated so=20 many visionaries: Risk.

By Robert O'Harrow Jr. and Brady Dennis
Washington = Post=20 Staff Writers
Monday, December 29, 2008; A01

First of three parts

Howard Sosin and Randy Rackson conceived their financial revolution = as they=20 walked along the Manhattan waterfront during lunchtime outings. They = refined=20 their ideas at late-night dinners and during breaks in their busy days = as=20 traders at the junk-bond firm of Drexel Burnham Lambert.

Sosin, a 35-year-old reserved finance scholar who had honed his = theories at=20 the famed Bell Labs, projected an aura of brilliance and fierce = determination.=20 Rackson, a 30-year-old soft-spoken computer wizard and art lover, = arrived on Wall Street with a Wharton School pedigree and a desire to create something = memorable.

They combined forces with Barry Goldman, a Drexel colleague with a = PhD in=20 economics and a genius for constructing complex financial transactions. = "Imagine=20 what we could do," Sosin would tell Rackson and Goldman as they = brainstormed in=20 the spring of 1986.

The three men had earned plenty of money through short-term deals = known as=20 interest-rate swaps, a clever transaction designed to protect banks,=20 corporations and other clients from swings in interest rates that threw=20 uncertainty into the cost of borrowing the money necessary for their = business=20 operations.

They believed their revolution could never happen if they stayed at = Drexel.=20 Swaps in those days typically lasted no longer than two or three years. = The trio=20 envisioned deals lasting decades that would lock in profits and manage = risks=20 with unprecedented precision. But the junk-bond firm's inferior credit = rating=20 sharply raised its borrowing costs, making it a dubious and risky = partner for=20 such long-term deals.

Sosin and his team needed the backing of a company with deep pockets, = a=20 burnished reputation and the very top credit rating, a Triple A = institution as=20 unlikely to default as the U.S. Treasury itself. One name topped their wish list = that fall:=20 American International Group, or AIG, the global = insurance=20 conglomerate considered one of the world's safest bets.

They would find a partner for their venture. They would create an = elegant and=20 powerful system that earned billions of dollars, operating in the seams = and gaps=20 of the market and federal regulation. They and their firm would alter = the way=20 Wall Street did business, particularly in the use of derivatives, and = eventually=20 test Washington's growing belief that capitalism could safely thrive = with little=20 oversight.

Then, they would watch in disbelief as their creation -- by then in = the hands=20 of others -- led to the most costly rescue of a private company in U.S. = history,=20 triggering a federal investigation into AIG's near-collapse and making = AIG=20 synonymous not with safety and security, but with risk and ruin.

Over the past two decades, their enterprise, AIG Financial Products, = evolved=20 into an indispensable aid to such investment banks as Goldman Sachs and Merrill Lynch, as well as governments, municipalities = and=20 corporations around the world. The firm developed innovative solutions = for its=20 clients, including new methods to free up cash, get rid of debt and = guard=20 against rising interest rates or currency fluctuations.

Financial Products unleashed techniques that others on Wall Street = rushed to=20 emulate, creating vast, interlocking deals that bound together financial = institutions in ways that no one fully understood and contributed to the = demise=20 of its parent company as a private enterprise. In the panic of = mid-September's=20 crash, the Bush administration said that AIG had grown too intertwined = with the=20 global economy to fail and made the extraordinary decision to take over = the=20 reeling giant. The bailout stands at $152 billion and counting -- almost = 10=20 times as large as the rescue for the American auto industry.

Many of the most compelling aspects of the economic cataclysm can be = seen=20 through the story of AIG and its Financial Products unit: the failure of = credit-rating firms, the absence of meaningful federal regulation, the = mistaken=20 belief that private contracts did not pose systemic risk, the veneration = of=20 computer models and quantitative analysis.

At the end, though, the story of Financial Products is not about math = and=20 financial formulas. It is a parable about people who thought they could = outwit=20 competitors and market forces alike, and who behaved as though they were = uniquely positioned to sidestep the disasters that had destroyed so many = financial dreams before them.

2: 'We Are The Tide'

Sosin, Rackson and Goldman could hardly contain themselves as they = labored=20 over a business plan at Sosin's kitchen table in his apartment on = Manhattan's=20 Upper East Side. Their timing happened to be exquisite. The staid Wall = Street of=20 their fathers' generation was gone, replaced by an anything-goes culture = that=20 applauded the kind of path they were charting during the final months of = 1986.

Their plan fit perfectly with another revolution they saw unfolding = in=20 Washington. Ronald Reagan's unwavering belief in free markets -- and = his=20 distaste for regulation that put hurdles in the way of entrepreneurs -- = had=20 steadily spread through the government. "The United States believes the = greatest=20 contribution we can make to world prosperity is the continued advocacy = of the=20 magic of the marketplace," Reagan told a U.N. audience that fall.

As eager as the three dreamers were, they had to confront certain = realities.=20 They had no backing, no inside track to the top levels of the corporate = world=20 that controlled the money they needed.

They had passed AIG headquarters at 70 Pine St., a few blocks from = Drexel's=20 offices, many times. Now, they wanted an entr=C3=A9e to the 18th floor, = where=20 legendary 61-year-old chairman and chief executive Maurice "Hank" = Greenberg=20 presided over the nation's largest insurance company, with operations in = scores=20 of countries. Greenberg was proud and protective of his company's AAA = credit=20 rating, one of only a handful in the world.

The AAA, awarded after an examination by the bond-rating firms, sent = a=20 resounding signal to clients that they could always sleep well at night, = that=20 AIG was in no danger of failing. The more secure a company, the more = cheaply it=20 could borrow money -- a fact that would be pivotal to Financial = Products'=20 success.

AIG's roots went back to 1919 and Shanghai, where founder Cornelius = V. Starr=20 built a business around a lucrative, relatively untapped insurance = market.=20 Starr's company later received an unorthodox boost when he worked with = the U.S.=20 Office of Strategic Services during World War II to create an = intelligence unit=20 that gleaned information from insurance documents.

When Greenberg took the reins in 1968, AIG was a privately held = company.=20 Greenberg, a compactly built son of a taxi cab driver, eventually became = a=20 figure in both New York and Washington, where he counted Henry Kissinger and Reagan CIA director Bill Casey among = his=20 confidantes. The World War II and Korean War veteran had a temper, a = gift for=20 growth and a restless mind. He had transformed AIG into a global titan = and now=20 wanted to do more.

Few people thought of AIG as a financial innovator. Greenberg kept = his=20 stockholders happy by striving for an annual 15 percent increase in = profits. He=20 instructed his deputy, vice chairman Edward E. Matthews, to explore how = AIG=20 could get more involved in Wall Street's realm.

"This is never going to get any better than it is today," Greenberg = told=20 Matthews. "We're so big, we're never going to swim against the tide. We = are the=20 tide."

3: 'It Wasn't The Money'

At the law offices of Kaye Scholer in Midtown Manhattan, former Sen. = Abraham=20 Ribicoff had a match to make.

Sosin had come to the firm -- where the 76-year-old Ribicoff was a = senior=20 adviser -- seeking guidance on how to leave Drexel. As he mentioned his = interest=20 in getting AIG's backing for a new venture, a Kaye Scholer lawyer told = him to=20 see Ribicoff, an old Greenberg friend.

Ribicoff was happy to introduce the inventive Sosin to the ambitious=20 Greenberg, and let them figure out whether they could do business = together. But=20 he warned Sosin that any partnership, no matter how productive, can = sour. "I'll=20 only call Greenberg if you let us plan your divorce while we're planning = your=20 marriage," Sosin remembers Ribicoff saying.

Sosin came to the negotiation with conditions. He wanted the kind of = autonomy=20 that Greenberg rarely granted. Greenberg wanted assurances that Sosin's = venture=20 would do nothing to harm the gold-plated rating he had spent two decades = building.

Greenberg had little extra time for the nuts-and-bolts details that = Sosin=20 sought to negotiate. "I don't really know much about this," he told = Matthews.=20 "You go talk to these people."

The morning after AIG and Sosin signed their joint venture agreement, = Jan.=20 27, 1987, word spread rapidly through Drexel's trading floor in lower = Manhattan:=20 Sosin, Rackson and Goldman were leaving. Discreetly, the three men had = invited=20 some of their colleagues to a recruitment meeting. Ten eventually signed = up for=20 the ride.

Michael Milken, the junk-bond king who was Drexel's star trader, = tried to=20 stop the breakaways. But the pull of innovation, and the promise of even = greater=20 pay, was too strong.

At Drexel, Sosin, Rackson and their band of brainy followers didn't = have much=20 say in how bonuses were doled out. At Financial Products, they would = keep 38=20 percent of the profits, with Greenberg and AIG getting 62 percent. = (Greenberg=20 remembers AIG's share as 65 percent.)

Their revolution began with a whisper. They set up shop in a = windowless,=20 makeshift room at an accounting firm on Third Avenue. Until the rental = furniture=20 arrived, they sat on cardboard boxes. When it finally showed up, someone = had=20 made a mistake and so for a short time, they perched on children's = chairs and=20 worked at tiny tables. When Matthews escorted Greenberg there for a = visit, the=20 chief executive chewed him out. "You can't have them in such terrible = quarters,"=20 Greenberg said.

Sosin and Rackson hoped that everyone would get rich, but they had = their=20 sights set on something more. They wanted to tear down walls they saw as = impediments to innovation, the "fiefdoms" that were standard practice at = other=20 Wall Street firms. Their vision required a collaborative culture and a = computer=20 system that no one else had. For six months, the group worked on = constructing=20 "the position analysis and storage system," or PASS. They called it = simply "the=20 system."

It enabled Financial Products to bring a rare discipline to complex = trades.=20 By maintaining market, accounting and transaction details in one place, = Sosin=20 and his people could track the constantly changing value of a trade's = components=20 in a way no other firm could.

Put more simply, they could see opportunities in the marketplace for = taking=20 on risk that others couldn't, squeeze out profits where no one had = before and=20 protect themselves in the process.

They exploited the developing realm of derivatives, financial jargon = for a=20 contract settling in the future that is based on something trading now. = A=20 futures contract is a common derivative: A farmer might agree to sell = wheat next=20 spring for a price set today. If the price goes up, the farmer misses = out on=20 greater profits; if it goes down, the farmer is protected against loss.=20 Essentially, the contract guarantees enough money to keep the farm = going.

For its clients, Financial Products found ways to create more = lucrative and=20 longer-term derivative deals tied to all sorts of underlying assets,=20 neutralizing the constant gyrations of prices in stocks, currencies and=20 commodities. Behind each transaction was the cushion of AIG's AAA = rating.

Precision was the key to tamping down the risk of these derivatives = to the=20 firm. Using another computer program to monitor the minute fluctuations = in=20 various rates, Financial Products could place offsetting trades on all = sides of=20 a transaction, so it almost didn't matter what the markets did. That was = the=20 beauty of their evolving machine: The firm won either way, as long as it = stuck=20 to its commitment to keep hedging its bets.

But it took more than technology to realize their vision. It took a = culture=20 of skepticism. The firm set up a committee to examine all transactions = at the=20 end of each workday, searching for flaws in logic, pricing and hedges. = "Everyone=20 kind of understood what the nature of the game was. . . . This was not a = company=20 that involved speculating," said Tom Savage, a mathematician from Drexel = who=20 joined the firm in 1988. "So it was everybody's job to criticize and=20 double-check other people's opinions about what was appropriate business = and=20 what wasn't."

Sosin and his colleagues worked to create a finely balanced system = that=20 married technology, intelligence, verve and cultural discipline.

"We were all kind of artists," Rackson said recently. "The excitement = of it=20 wasn't the money. The money was the scorecard. The drive behind it was = creating=20 something new."

4: 'We Regret to Inform You . . .'

In July 1987, Sosin phoned Ed Matthews at his vacation house in the=20 Adirondacks, where the AIG executive often went to escape Manhattan's = summer=20 heat. It was a phone call both would remember for a long time.

Financial Products was about to close its first significant deal, a = $1=20 billion interest-rate swap with the Italian government, 10 times larger = than the=20 typical Wall Street swaps deal in those days.

The elements of the transaction might seem arcane to those outside = the=20 financial world. The contract involved an exchange of floating and fixed = rates=20 that gave Italy advantages in how it paid bondholders. Financial = Products=20 engaged in a separate set of transactions to offset the risk it was = taking on.=20 As Sosin explained to Matthews, the firm made money, over the life of = the=20 contract, on the spread between the cost of the deal and the cost of its = hedge.

This one swap, Sosin told him, would pay the firm more than $3 = million -- as=20 much as AIG's two other small financial operations each earned in a = year.

"I was stunned," Matthews said.

That first year, Financial Products brought in millions for the = company --=20 $60 million in the first six months alone, as Sosin recalls. He and his = team=20 left behind their ad hoc digs for a swanky Madison Avenue address, a = temporary=20 stop en route to their eventual headquarters in suburban = Connecticut.

Competitors hustled to keep pace. Sosin pressed to find niches where = others=20 weren't playing and provide cost-saving solutions for clients. Standard=20 interest-rate swaps were no longer enough. The firm moved into more = exotic=20 deals, involving stocks, currency and municipal bonds.

By 1990, Financial Products had offices in London and Tokyo. It would = soon=20 set up a small bank in Paris to improve its image and lower the cost of = some=20 European deals.

As in the Italian deal, the transactions were hedged and, if = necessary,=20 hedged again. The hedges involved precisely calibrated transactions, = including=20 the purchase of Treasury bonds or other swaps, that brought a cash flow = in=20 almost direct proportion to the money going out.

But with success came tension. Greenberg's love of his joint = venture's=20 revenue could not overcome his desire for greater control. He chafed at = the=20 deal, worrying that he had given Sosin too much freedom.

One detail in particular nagged at Greenberg. Under the joint-venture = agreement, Financial Products received its profits upfront, even if the=20 transactions took 30 years to play out. AIG would be on the hook if = something=20 went wrong down the road, not Sosin and his team, who took their pay=20 immediately.

Greenberg's uneasiness grew into distrust, and not just about the = numbers.=20 Greenberg was a wink-and-handshake guy, while Sosin relied on the = written=20 agreement as his Bible. If Greenberg asked for something that wasn't = stipulated,=20 Sosin wouldn't comply.

"We ran our company very openly," Greenberg said. "Our word was our=20 bond."

For his part, Sosin said the agreement gave both sides a clear = understanding=20 of the arrangement.

Early in 1990, Greenberg summoned Sosin to his office. Drexel had = just=20 imploded amid allegations of fraud and insider trading, and Greenberg = had=20 recruited several executives to start an AIG unit specializing in = currency=20 trading. That was a problem: Sosin interpreted the joint agreement as = giving his=20 firm exclusive rights to that business. Greenberg disagreed, and hoped = to=20 finesse the conflict.

"Howard, I'm sure you won't mind," Greenberg said.

"Mr. Greenberg, I mind very much," Sosin said.

"Howard, that isn't wise," Greenberg responded.

Days later, on March 13, 1990, Matthews sent Sosin a letter on = Greenberg's=20 behalf announcing their intention to terminate the agreement. "We regret = to=20 inform you. . . " the letter began.

Under the agreement, Sosin could take a duplicate of his computer = system and=20 his team with him. He began looking for backing from another AAA = company.=20 Greenberg heard about Sosin's efforts and got cold feet. After a series = of=20 meetings, including one at Greenberg's Florida retreat in Ocean Reef, = they=20 patched it back together, reasoning that there was too much money still = to be=20 made.

Greenberg's next letter had a different tone. "It is with great = pleasure=20 that, with this letter, we revoke any and all of our prior notices of=20 termination," he wrote on May 31, 1990.

The peace wouldn't last.

5: 'Cave or Terminate'

In late 1992, Greenberg once again summoned Sosin to AIG = headquarters. He was=20 livid over two recent Financial Products deals with entities controlled = by the=20 Edper Group, a giant Canadian holding company owned by billionaires = Edward and=20 Peter Bronfman.

The first involved the purchase of bonds, which amounted to a loan to = one of=20 the Edper entities. The firm occasionally ventured into such credit = deals as=20 part of larger transactions, but only with highly rated companies and = with=20 provisions that opened an exit ramp if the bonds started to default. "We = want to=20 be the first rat to leave the sinking ship," Sosin told his troops, = reflecting=20 his unease with credit deals, which their system couldn't tame.

When this particular ship sank, Financial Products sold out as = quickly as it=20 could, but not before it lost $100 million. The second deal, involving a = swap=20 with extra layers of complexity, was going fine. But the $100 million = loss in=20 the first deal and the intricate machinations in the other had spooked=20 Greenberg.

Sitting in an anteroom to his office, in a favorite red leather = chair,=20 Greenberg demanded that Sosin stop doing some of the deals that had made = Financial Products a Wall Street darling.

Greenberg handed Sosin a document that would change the terms of = their joint=20 venture. Greenberg was daring Sosin to flinch. Instead, Sosin walked = out.

He visited his lawyer, Ronald Rolfe, at Cravath Swaine & Moore in = New=20 York.

"I said, 'What can I do?' And he said, "Cave or terminate.' = "

6: No=20 Reconciliation Possible

Under the agreement, either man had the right to terminate the joint = venture.=20 Sosin notified Greenberg that he wanted out.

Greenberg knew that Sosin's departure could cost him and AIG = millions. But=20 that wasn't his main concern. He didn't have a thorough understanding of = how=20 Sosin's system worked, and he wasn't going to let him get away without = finding=20 out.

In March 1993, as the two sides commenced a bitter arbitration = battle,=20 Greenberg formed what came to be known as a "shadow group." It verged on = a=20 covert operation. The group included AIG's auditors, now known as PricewaterhouseCoopers, which set up an office near = Financial=20 Products -- now in Connecticut -- and built a parallel computer system = to track=20 the firm's trades. Greenberg also held surreptitious conversations with = some of=20 Sosin's colleagues, recruiting them to stay.

Years later, Greenberg and Matthews still chafe visibly at the = mention of=20 Sosin. "One of the most difficult individuals I have ever dealt with in = my=20 entire life. Hands down," Matthews said. "Howard was in it for = Howard."

Sosin, too, remains sensitive about what happened. "Greenberg took = this very=20 personally," he said. "He likes to be able to step in at any point and = change=20 things at his whim."

In Sosin's view, Greenberg and Matthews were envious of the profits = that he=20 and his colleagues were keeping for themselves. "It was peculiar to have = something go so well," Sosin said, "and for him to have such = suspicion."

In August 1993, with no reconciliation possible this time, the AIG = board of=20 directors installed a new leadership team. Sosin and Rackson took some = employees=20 with them to start another firm. Sosin later settled with AIG for a = reported=20 payout of more than $150 million; Rackson later received a share of the=20 settlement.

Greenberg and AIG gained control of Financial Products and the = beautiful=20 machine. In the coming years, the firm would accelerate its = profit-making=20 ability, while forging into uncharted -- and ever riskier -- financial=20 territory.

7: 'Honor the Trust'

Tom Savage stood before a room of anxious colleagues at the Four = Seasons=20 resort in Dallas, eager to reassure them that Greenberg was not going to = pull=20 the plug on their money-making machine.

Savage, a 44-year-old Midwestern math whiz, had just been named the = new=20 president of Financial Products. With the honor came explicit = expectations,=20 which Greenberg made clear: "You guys up at FP ever do anything to my = Triple A=20 rating, and I'm coming after you with a pitchfork."

It was spring 1994 and, on the surface, nothing much had changed = since Sosin=20 left the previous summer. Financial Products had come a long way from = the days=20 of sitting on cardboard boxes. The Dallas meeting was opulent in a way = that had=20 become customary for the firm: Lavish meals, open bars, luxurious rooms = and=20 rounds of golf, which was Savage's particular passion. Dallas's = international=20 airport allowed dozens of associates to fly direct from the firm's = far-flung=20 outposts.

The employees couldn't understand why there was any doubt about the = firm's=20 future. In just seven years, it had grown into a 125-person operation = with=20 annual profits comfortably above $100 million.

Like his predecessors, Savage knew the enterprise could not thrive = without=20 AIG's AAA rating, which continued to provide the leverage it needed to = stay=20 ahead.

"AIG has given us the license to work," Savage told his colleagues = that day.=20 "We have to honor the trust they have given us."

The catch? Financial Products would have to take more direction than = ever=20 from Greenberg.

8: 24 Hours A Day

Greenberg called Savage most days that first year. "I'd be changing a = diaper=20 at home," Savage recalled. "He'd say, 'What are you doing?' I'd say, = 'Changing=20 the diaper.' He'd say, 'Well, I don't think I can help you with that.' = But he=20 would say, 'What are you thinking about? What's going on?' He was always = taking=20 my temperature."

Savage knew that Greenberg hadn't been 100 percent sure about his = ability to=20 run Financial Products. Greenberg had told him as much when they sealed = the deal=20 at a Vermont ski resort that AIG owned in Stowe. "I don't know if you = have all=20 the buttons for this job," the AIG chairman had said. Greenberg managed = the=20 company by both charming and intimidating his subordinates. He said of = himself=20 recently, "I suffer fools very badly."

Greenberg also had no patience for anyone who didn't share his = relentless=20 work ethic. "You don't build a company like AIG from nine to five, five = days a=20 week. It just doesn't happen," Greenberg said recently. "And you've got = to=20 surround yourself with a group of people who share the same values, the = same=20 aspirations that you do. When I traveled, I could call somebody, I don't = care=20 what time it was, maybe two, three in the morning. As far as I'm = concerned, I'm=20 working 24 hours, they're working 24 hours."

Savage understood that, but he came at the job with a mathematician's = love of=20 the numbers and how they worked. He was among a growing number of = "quants" --=20 short for quantitative thinkers -- who had worked their way into the = heart of=20 Wall Street. With a PhD from Claremont Graduate University in = California, Savage=20 had started his career at First Boston in 1983, where he wrote computer = models=20 for a then-arcane type of security called a collateralized mortgage = obligation,=20 or CMO. It is the kind of asset-backed security at the core of the = current=20 meltdown.

Savage respected Sosin, but saw no reason to follow Sosin and Rackson = out the=20 door. "I think what was clear was that, however things should work out, = there=20 was a business at AIG Financial Products and Sosin didn't need to be = there for=20 it to be successful," Savage said.

Immediately after Sosin's ouster, Savage and three others -- soon = dubbed the=20 Gang of Four -- ran Financial Products on an interim basis, with = Matthews=20 assigned to keep tabs on them. Savage remained committed to running the = place=20 under the same rigorous, risk-reducing code that Sosin's group had=20 cultivated.

But not everything stayed the same. Under a new operating agreement = imposed=20 by Greenberg, AIG owned Financial Products as a subsidiary, and the = parent=20 company received 70 percent of the profits, up from 62 percent.

Greenberg also wanted to change the way Financial Products' employees = divvied=20 up its share of the profits. Under the previous arrangement, Sosin and = his crew=20 had the right to book immediate profits on the long-term deals. = Greenberg=20 thought there was a powerful incentive to go after millions of dollars = in=20 short-term gains while leaving AIG and its shareholders responsible for=20 potential losses for years to come.

Savage agreed with Greenberg that Financial Products employees should = defer=20 half of their compensation for several years, depending on the length of = the=20 deals being done -- an arrangement that would still yield hefty = paychecks as the=20 firm's profits soared in the coming years.

Savage said he welcomed Greenberg's input. "I would give Greenberg a = lot of=20 deference," Savage said. "Hank Greenberg's a great man. And I'm willing, when I = talk to him,=20 to say, you know, I'm in the presence of a great man and that's worth=20 something."

9: 'We're Not Hiding Anything'

Financial Products found its profit margins shrinking on some = transactions as=20 competitors succeeded in duplicating its services. Like Sosin, Savage = urged his=20 talented team to devise ever more complicated transactions, often in = untapped=20 areas.

Financial Products was becoming a chameleon, taking on the coloration = of=20 whatever problem it was solving for its diverse clients. The firm pushed = further=20 into structured investments, hedge fund deals and guaranteed investment=20 contracts, or GICs. The GIC deals involved loans from municipalities = that had=20 temporary surpluses of cash. Financial Products reckoned that it could = borrow=20 that cash, pay state and local governments more than they could make = otherwise=20 and then use the money for lucrative deals for itself, somewhat like a = bank.

The firm also began applying its complex formulas to the movement of = single=20 stocks. Using such structured finance enabled clients, such as Microsoft, to better manage their stock prices. It also = helped=20 Financial Products to more than double its profits in three years -- to = $323=20 million in 1998, from $140 million in 1995.

A new unit, called the Transaction Development Group, did its part by = taking=20 advantage of gaps between securities regulation and tax laws in the = United=20 States as well as in other countries. Financial Products associates = noticed, for=20 instance, they could make money by exploiting differences between the = U.S. and=20 British definitions of stocks and bonds. A security that met the = definition of=20 stock in Britain could pay tax-free dividends to shareholders. The same = security=20 in the United States was regarded as a bond that provided tax-deductible = payments. A Financial Products client would get both tax breaks. The = firm used=20 the capital raised from that line of business, in part, to finance other = operations.

"We're the guys there who are going to try to exploit that," Savage = said. "We=20 dot our i's, we cross our t's, we tell everybody what we're doing. We're = not=20 hiding anything. . . . However, we're getting different treatments in = different=20 jurisdictions and we're making money as a result."

But even as Financial Products experimented, Savage said, he = continued to=20 stress the need to minimize risk. "That was one of the things that = really marked=20 this company, was the rigor with which it looked at the business of = trading. . .=20 . There was an academic rigor to it that very few companies match," he = said.

"It was Howard Sosin who said, 'You know, we're not going to do = trades that=20 we can't correctly model, value, provide hedges for and account for.' = "

Though the language of caution was the same, the firm's drive toward = novel=20 and ever more lucrative deals led down the path of greater risk. The = beautiful=20 machine was about to crack.

Tuesday: The dangerous fork in the road

Staff writer Bob Woodward contributed to this report.