(FORTUNE Magazine) – There is something deeply compelling about a comeback story: Lance Armstrong beating back cancer to win the Tour de France, Ali beating back Foreman in Zaire at the ripe age of 32, Tina Turner beating back Ike. Such tales tug at our emotional core, satisfying the human craving for long shots and second chances and movies of the week.

No wonder the story of Oxford Health Plans has been so thoroughly embraced by Wall Street. Not long after its spectacular stock market flameout in October 1997--in which Oxford saw $3.4 billion in market value vanish in a single afternoon--the health insurer was considered a lost cause by most investors, securities analysts, the news media (which delivered more than a few scathing postmortems), and especially the doctors in its network. After a botched conversion of its computer system and the news that Oxford suddenly owed millions of dollars in back claims to medical providers, the stock began its precipitous fall from a high of $88 a share to $6 in August 1998.

That's when Norm Payson arrived. A rumpled former family practitioner who had worked on an Indian reservation soon after med school, Payson seemed as if he'd been sent from central casting. Right away the gentle-spoken new CEO knew what to do. Within months, he had shut down unprofitable lines of business such as Medicaid and certain Medicare plans. He closed several offices, laid off employees, fixed the star-crossed computer system, and raised much-needed cash from private investors. Wall Street's response was encouraging from the start. With several analysts showering accolades on the bold new boss, the stock climbed steadily over the next two-plus years from $6 a share to its current $32--representing a gain that's more than 80 times that of Standard & Poor's health care/managed care index.

Indeed, investment analysts have been ebullient in applauding their comeback kid. "The turnaround is over," proclaimed UBS Warburg analyst William McKeever, in a recent refrain that echoes that of researchers at Prudential and Morgan Stanley Dean Witter. "The company was teetering on the abyss.... They have righted the ship. The company's back." Thirteen of the 15 analysts who cover the Trumbull, Conn., company now rate it a buy. Nor is Payson himself shy about touting Oxford's achievement. In the very first line of 1999's annual report, the CEO declared that "Oxford effected one of the greatest financial and operational turnarounds...in the history of the industry."

There's just one problem with this feel-good scenario: It doesn't appear to be true.

For one thing, the explosive earnings growth that has won Oxford such lavish praise on Wall Street over the past two years has come, in part, from the release of excess reserves intended for future medical claims--reserves that regulators had ordered Oxford to build up around the time of the 1997 crash. Second, there's considerable evidence that the insurer, once among health care's most generous companies, has been delaying claims and skimping on coverage to reduce medical expenses.

Now that Oxford has spent months wringing savings out of everything from administrative overhead to medical provider reimbursements, it will have to begin increasing earnings the old-fashioned way: by growing the business. But since Oxford's membership rolls have long been moving in the opposite direction, that is unlikely to come easy. The HMO's customer base shrank 29% from the first quarter of 1998 through the third quarter of 2000--from around 2.1 million members to fewer than 1.5 million. The cause wasn't merely the company's deliberate departure from Medicare and Medicaid. In Oxford's three commercial plans, membership dropped nearly 21% for the period, slipping in each consecutive quarter. While the trend reversed in the final quarter of last year, with overall membership creeping up 0.4%, the company has been lagging far behind the industry's leaders.

Oxford is confident it can accelerate its growth. During the company's fourth-quarter earnings conference call, Payson hailed the recent uptick in membership as "the beginning of a new era of growth." He forecast that commercial membership would grow 5% this year. Payson also estimated that Oxford would make $4.4 billion this year and earn $2.60 a share. Several analysts, though, have already priced a more dramatic recovery into the stock. Morgan Stanley's Christine Arnold, for instance, forecasts a 9% bump in commercial membership, and per share earnings of $2.70.

As for how Oxford will get there, the path isn't exactly clear. Its latest growth strategy--enrolling small businesses in its new Metro plan--seems a dismal failure. What's more, there's strong evidence to suggest that Oxford's prized physician network--along with its once pristine reputation among doctors--is badly deteriorating. Many physicians are now downright furious with the company that always promised to be a different kind of HMO. The bitterness may well make it hard for Oxford to keep providers and patients in the organization, let alone attract new ones.

But what is perhaps most remarkable about the seeming disconnect between Oxford's difficulties and Wall Street's rosy expectations is that we have been here before. In a big way.

There was a time not long ago when Oxford was one of the fastest-growing health-care companies in history. Its membership rolls soared from 117,000 customers in 1992 to two million in 1997--a five-year growth rate of more than 1,600%. Revenues over the same period exploded, doubling nearly five times, from $151 million to $4.2 billion. And then there was the stock: After Oxford went public at a split-adjusted $2 a share in 1991, the stock shot up to $88 in 1997. For the five years ended 1996, Oxford had the second-highest shareholder return on investment of all publicly traded companies (85% annually).

And then it crashed--with its stock dropping 62% in one day. Few people saw it coming in 1997. Should they see it now?

In an industry in which many key financials are guarded as closely as state secrets were during the Cold War, one figure has been heralded by analysts as evidence of Oxford's resurgence: its "medical-loss ratio." This figure represents the amount of money an HMO pays out in medical care divided by the premiums it receives. A medical-loss ratio of 82%, which means that the company has paid 82 cents of every dollar in premium revenue on health-care costs, is considered quite good. The lower the ratio, the better the profits and earnings.

The general industry range is between 82% and 86%. Then there is Oxford, which finished last quarter with a mind-bogglingly attractive medical-loss ratio of 72.5%. That ratio, says Merrill Lynch analyst Roberta Goodman, has pushed the company's operating margins to just over 11%, tops in the industry. Oxford, in fact, is twice as profitable as the average HMO, Goodman points out.

Oxford officials, such as CFO Kurt Thompson, are quick to credit Payson's cost-management measures for the company's extraordinary benchmark. Among other things, he negotiated long-term contracts with doctors and other providers such as pharmacies and radiologists, which markedly reduced the per claim cost of care. He also started "profiling," or monitoring, physicians who appeared to be charging outside the norms, in order to bring them more in line with their peers.

On the premium side, Payson gambled that employers would pay more for the insurer's well-regarded brand, and raised premium yields--the net effect of premium increases and benefit reductions--7.3% a year between 1998 and 2000. During that three-year span, Oxford's premiums rose faster than its medical costs. The difference went directly to the bottom line.

But as the company's financial statements reveal, there may also be a more troubling reason for Oxford's low loss ratio. The insurer has been steadily releasing money from its reserve for future medical claims. Specifically, it has been letting money out of a reserve for what's known as incurred-but-not-reported claims, or IBNR. In HMO accounting, IBNR is an actuarial estimate of medical claims that plan members have incurred (for doctor's visits, diagnostic tests, surgical procedures, etc.) but that have not hit the books because the provider or member has not yet reported the claim to the insurer. Some have expressed concern that Oxford is feeding off its reserve, which is meant to pay for patients' medical care. But on Wall Street, there are few raised eyebrows. Instead, the practice is considered more evidence of Payson's magical touch. "I don't know of any other HMO that has consistently been able to release its reserves," marvels analyst Robert Mains of Advest, a financial advisory firm in Hartford, Conn., who rates Oxford's stock a buy. He says releases from reserves demonstrate how impressive Oxford is as a "medical-cost-containment story" since they indicate that the company's medical expenses have been lower than it had estimated.

But here's the problem: The amount that's released reduces the "health-care expense" component in calculating the medical-loss ratio, making that MLR look better. In translation, Oxford's cost of care looks cheaper than it actually is. It's a fine tactic, but one that works for only a limited time.

Take last year's third quarter. Factoring in a $30 million release from the IBNR claims reserve, Oxford reported earnings per share of 81 cents--nearly triple the amount from the year-previous period--on an unthinkably low medical-loss ratio of around 75%. Excluding the reserve adjustment, Oxford's loss ratio would have been 78%, with per share earnings falling to 65 cents. (Other factors, as we'll discuss, are likely to move the ratio up more sharply.)

While the company always discloses the impact the reserve release had on reported results, clearly not everyone picks up on it. Perhaps because Oxford and most of the Street's analysts focus mainly on the adjusted ratio, popular stock-research houses such as Standard & Poor's, and much of the media, reported only the better earnings number. In response to the third quarter's impressive performance, investors bid up Oxford shares in the ensuing six weeks to $42, a 56% gain.

None of this suggests that Oxford is doing anything illegal. Because inadequate IBNR claims reserves were one of many problems that contributed to the company's October 1997 meltdown, Oxford was required in 1997 by state insurance regulators to boost this reserve substantially; at the end of the year it had reached $937 million. As Payson and his team have worked through an enormous backlog of medical claims, exited high-loss businesses like Medicaid, and improved cost controls, Oxford has been able to reduce this liability substantially, says CFO Thompson. Releasing excess reserves falls within Generally Accepted Accounting Principles, and Oxford's auditors and state regulators sign off on the IBNR reserve estimates annually.

Nevertheless, the adjusted loss-ratios exaggerate the strength of Oxford's core business. And now that the company has reduced its IBNR claims reserve to $519 million--releasing $135 million directly into earnings over the past six quarters--it may not have much more that it can draw. Since Oxford is one of the few HMOs that reports IBNR claims reserve separately from medical claims payable, it's difficult to compare its reserve for IBNR claims with those of its peers. But the two analysts who raised questions about Oxford before its 1997 crash--Christopher Teeters at the Center for Financial Research and Analysis, an independent research firm in Rockville, Md., and Anne Anderson, president of Atlantis Research in Parsippany, N.J.--suspect that Oxford's IBNR claims reserve may now be about as low as it can go. Oxford refuses to say how low this reserve can fall, and declines to offer a range at which it might stabilize.

The insurer seems to have brought down its loss ratio in yet another way. Evidence suggests that in some cases, the company has been trying to avoid paying its medical bills. Critics, in fact, say much of Oxford's legendary turnaround has come at the expense of providers. State and county medical societies in New York, New Jersey, and Connecticut report that Oxford's late and denied payments routinely top their lists of physician complaints against Oxford. A common technique, say doctors, is to deny claims by stamping them as "unclean"--that is, declaring that not all the paperwork is in order. Yet when physicians ask what the company is looking for, the insurer refuses to clarify, these doctors maintain. "It's a very passive way to deny claims," says Robert Goldberg, the immediate past president and a current board member of the New York County Medical Society.

Oxford spokeswoman Maria Gordon-Shydlo responds that it is "far simpler and less expensive" to pay claims as they come in, so the company would never classify clean claims as unclean in order to delay processing. She adds that only 4% to 5% of Oxford's 60,000 daily claims are "dirty," which she says is about in line with the industry average. Still, Oxford's cost-cutting measures have been attracting the attention of regulators, who have recently been cracking down hard. In New Jersey the company was fined a record $275,000 last March for failing to pay claims promptly. While the fine covered services that were provided during Oxford's troubled years in 1997 and 1998, it also related to those administered in 1999--a year in which Oxford officials insist the company was back on its feet. In New York, the insurance department fined 21 health insurers a total of $575,000 last October for prompt-payment violations during the fourth quarter of 1999; the largest fine, $215,000, was levied against Oxford.

Oxford is hardly the only offender--such abuses seem almost endemic to a hard-pressed HMO industry. But it now looks as though the company may be among the worst.

And it goes beyond playing hardball with doctors. Some plan members may be at risk as well. A group of advocacy organizations led by New York City's public advocate went so far as to issue a 57-page investigative report titled "How Oxford Is Compromising the Health of Chronically Ill Patients." The report, released in November 1999, warned that some 50,000 Oxford members who purchased their health insurance directly--that is, not through an employer or public programs like Medicare and Medicaid--could find themselves suddenly without their chosen doctor after being transferred automatically to a different, and more bare-bones, health plan. Still, Oxford went ahead with the switch.

Oxford's Gordon-Shydlo says the company transferred these patients in response to members' desire for lower premiums and that the vast majority (84%) were able to maintain their previous providers. She adds that the New York State Department of Insurance and the Department of Health approved the transfer.

In Connecticut, the story is much the same. The state medical society, representing 7,000 doctors, filed suit last month against Oxford and five other HMOs, claiming the companies arbitrarily denied crucial medical treatment and illegally withheld millions of dollars in payments to doctors. If you buy into the notion that Oxford's executives have turned the troubled firm around, says Mark Thompson, executive director of the Fairfield County Medical Association in Trumbull, Conn., it's for one simple reason: "They did it on the backs of providers."

Avital Fast, chairman of the department of rehabilitation medicine at New York's Montefiore Medical Center, fervently agrees. Montefiore stopped accepting Oxford patients for physical therapy within the last year, after the HMO required that an outside firm "pre-certify" all prescriptions--a task that dramatically increases the already burdensome paperwork, Fast says. While the physician is no big fan of any HMO, he harbors particularly ill feelings for Oxford. "They really don't care for the patient as much as the bottom line," he fumes. "I think Oxford lost it. And I am really pissed!"

Oxford vigorously denies that ties to doctors are deteriorating. Spokeswoman Gordon-Shydlo adds that while 4% of doctors typically disenroll from the network each year, only 0.3% do so due to "stated dissatisfaction."

Even so, mindful of higher medical costs in the coming year--and perhaps the need for fence mending with doctors--Payson now thinks the company's 2001 medical-loss ratio is likely to be closer to 80% or 80.5%, as he told analysts during a February conference call. Truly repairing relations with doctors, though, may well boost Oxford's medical-loss ratio much higher than that.

While all of this cost cutting has no doubt helped juice Oxford's earnings growth, it certainly hasn't helped the insurer rebuild its business with providers and patients. The overall number of physicians in Oxford's network has risen between 4% and 10% annually over the past three years while competitors such as United Healthcare and Empire BlueCross BlueShield have seen physician enrollment surge. Empire, for example, which has tripled its physician enrollment in the New York metropolitan area in the past four years, now has more doctors in the region than Oxford has in its entire network. Likewise, on the membership front, there is little doubt the company is losing ground to competitors. For example, as United and Aetna increased rolls in the region by about 10% each during the 12 months ended June 2000, Oxford's membership fell 2%, according to UBS Warburg. Since 1997 Oxford's annual sales across the board have been flat, at about $4 billion.

Ultimately, however, Wall Street continues to have huge expectations that the company will find an earnings boost not from shaving more costs but from top-line growth. So what's up Payson's sleeve? Mainly, the rollout of a new health plan called Freedom Plan Metro. Launched in September, the new product targets a niche market of small-business owners. Employer premiums are 5% to 10% cheaper than those of rival small-group plans, but the costs are shifted to employees, whose co-payments will double or even triple in some cases. As for going out of network, the annual deductible is a steep $1,000 per individual and $3,000 per family.

Several analysts, somehow, seem convinced that the new plan will be a home run with small employers. The consensus estimate is that Metro will help Oxford boost earnings this year by 29%, more than triple the industry's projected growth rate of 8%. It's a familiar refrain, of course. At the beginning of 2000, analysts such as Morgan Stanley's Christine Arnold predicted that Oxford would jump-start its membership that year. Instead, even factoring in the tiny surge in 2000's fourth quarter, membership dropped 6% for the year.

For Oxford, though, substantial growth isn't a luxury, it's a necessity. The managed-care industry is brutally competitive. Until last year, when the trend eased, the number of HMO failures had been rising for five consecutive years, according to Weiss Ratings, an independent provider of ratings and analysis on the insurance industry. An oversaturated market, explains chairman Martin Weiss, has led to cutthroat pricing.

Though conditions are better now than they were a year or two ago, shareholders may not wait much longer for Oxford to break out of its membership doldrums. While the stock's price/earnings ratio is roughly in line with its peers', by at least one measure Oxford trades at an extraordinary premium: The ratio of its market capitalization to the number of its members is roughly 1,850--two to five times higher than those of its primary competitors.

Unfortunately for shareholders, Payson isn't signaling the kind of outlook for the Metro plan that infuses confidence. When asked how the new offering was selling during the company's year-end conference call in February, Payson declined to reveal any results. "It's not significantly in the numbers," he said. He also maintained that Metro is "building up steam," adding that if the economy softens and the market becomes more price sensitive "it will do disproportionately well."

Interviews with several insurance brokers, though, are clearly less sanguine. "People don't want to switch," says Andy Iversen, a director at health insurance wholesaler Benefitport, in Greenwich, Conn. "It's too expensive." Insurance broker Patrice Goldfarb agrees. "I've shown it to a couple of dozen employers," says Goldfarb, who is also president of the New York State chapter of the National Association of Health Underwriters, the nation's largest trade group for health insurance agents and brokers. "But I haven't had one person say, 'Yeah, this makes sense. Let's do it.' " She adds that the employers didn't think the savings were worth the aggravation they'd get from the employees for the steep out-of-pocket expense.

But perhaps the biggest problem of all may be a question of image. Once one of the most well-regarded companies in health care, routinely topping quality-of-care surveys, Oxford may have lost its most valuable asset for growing anew: its very cachet.

"There was a time when you'd go into a client's office and they'd ask for Oxford," sighs Goldfarb. "That doesn't happen anymore."

FEEDBACK: dstires@fortunemail.com