"MAN ON A BENCH," BY DUANE HANSON, 1977, POLYVINYL POLYCHROMED IN OIL, AND MIXED MEDIA WITH ACCESSORIES, PHOTOGRAPH COPYRIGHT THE SAATCHI GALLERY, LONDON ART COPYRIGHT ESTATE OF DUANE HANSON/LICENSED BY VEGA, NEW YORK

I.......... When I caught up with Robert S. Miller, the chief executive of Delphi Corporation, last summer, he was still pitching the fantasy that his company, a huge auto-parts maker, would be able to cut a deal with its workers and avoid filing for bankruptcy protection. But he acknowledged that Delphi faced one perhaps insuperable hurdle - not the current conditions in the auto business so much as the legacy of the pension promises that Delphi committed to many decades ago, when it was part of General Motors. This was the same fear that had obsessed Alfred P. Sloan Jr., the storied president of G.M., who warned way back in the 1940's that pensions and like benefits would be "extravagant beyond reason." But under pressure from the United Auto Workers union, he granted them. And as future auto executives would discover, pension obligations are - outside of bankruptcy, anyway - virtually impossible to unload. Unlike wages or health benefits, pension benefits cannot be cut. Unlike other contracts, which might be renegotiated as business conditions change, pension commitments are forever. And given the exigencies of the labor market, they tend to be steadily improved upon, at least when times are good.

For the U.A.W., Miller noted forlornly, "30 and Out" - 30 years to retirement - became a rallying cry. Eventually, the union got what it wanted, and workers who started on the assembly line after high school found they could retire by their early 50's. "These pensions were created when we all used to work until age 70 and then poop out at 72," Miller told me. "Now if you live past 80, a not-uncommon demographic, you're going to be taking benefits for longer than you are working. That social contract is under severe pressure."

Earlier this month, Miller and Delphi gave in to the pressure and sought protection under the bankruptcy code - the largest such filing ever in the auto industry. It followed by a few weeks the Chapter 11 filings of Delta Air Lines and Northwest Airlines, whose pension promises to workers exceeded the assets in their pension funds by an estimated $16 billion. The three filings have blown the lid off America's latest, if long-simmering, financial debacle. It is not hedge funds or the real-estate bubble - it is the pension system, both public and private. And it is broken.

The amount of underfunding in corporate pension plans totals a staggering $450 billion. Part of that liability is attributable to otherwise healthy corporations that will most likely, in time, make good on their obligations. But the plans of the companies that fail will become the responsibility of the government's pension insurer, the Pension Benefit Guaranty Corporation. The P.B.G.C., which collects premiums from corporations and, in theory, is supposed to be self-financing, is deeply in the hole, prompting comparisons to the savings-and-loan fiasco of the 1980's. Just as S. & L.'s of that era took foolish risks in part because their deposits were insured, the P.B.G.C.'s guarantee encouraged managements and unions to raise benefits ever higher.

In such situations, individuals are tempted to take more risk than is healthy for the group; economists, in a glum appraisal of human nature, call it "moral hazard." In effect, America's pension system has been a laboratory demonstration of moral hazard in which the insurance may end up bankrupting the system it was intended to save. Given that pension promises do not come due for years, it is hardly surprising that corporate executives and state legislators have found it easier to pay off unions with benefits tomorrow rather than with wages today. Since the benefits were insured, union leaders did not much care if the obligations proved excessive. During the previous decade especially, when it seemed that every pension promise could be fulfilled by a rising stock market, employers either recklessly overpromised or recklessly underprovided - or both - for the commitments they made.

The P.B.G.C. is now $23 billion in the red - a deficit that is expected to grow, significantly, as more companies go under. The balance sheet for the end of September will very likely show a deficit of more than $30 billion. If nothing is done to fix the system, the Congressional Budget Office forecasts, the deficit will mushroom to more than $100 billion within two decades. This liability will almost certainly fall back on the taxpayers, since the alternative to a bailout - letting the pension agency fail - would force aging former auto workers and other retirees onto the street.

As bad as that sounds, the problem of state and local government pensions is even worse. Public pensions, which are paid by taxpayers and thus enjoy an implicit form of insurance, are underfunded by a total of at least $300 billion and arguably much more. While governments have been winking at these deficits for years, they are now becoming intolerable burdens for taxpayers. In San Diego, pension abuse has effectively bankrupted the city. Thanks to a history of granting sweeter and sweeter pension deals that it has neglected to fund, the city has been forced to allocate $160 million, or 8 percent of the municipal budget, to the San Diego City Employees Retirement System this year, with similar allocations expected for years to come. San Diego has tabled plans for a downtown library, cut back the hours on swimming pools, gutted the parks and recreation budget, canceled needed water and sewer projects and fallen behind on potholes.

State or local governments in New Jersey, New York, Illinois, Ohio, West Virginia and elsewhere face similar budget strains aggravated by runaway pension promises. According to Carl DeMaio, director of the Performance Institute, which advocates better government accountability, "There is a San Diego brewing in every community."

Not only are taxpayers certain to suffer, but senior citizens in the future may also have to settle for less secure retirements anchored only by Social Security and whatever they've managed to put away into their 401(k) accounts. A backlash already has begun in state capitals, where the political forces that have been lobbying for Social Security reform have been rallying lawmakers to get out of the pension business altogether. Alaska's Legislature recently passed a shotgun bill to deny pensions to future employees. This mimics a trend in the private sector, in which corporations have been leaving the system, either by paying off their workers and terminating their pension plans or by "freezing" their plans, a step recently taken by Hewlett Packard, so that many current employees will no longer accrue benefits and new employees will not participate at all.

If the pension system continues to wither, it is not hard to envision a darker future in which - as was true until early in the 20th Century, before the advent of pensions - many of the elderly would be forced to keep working to stave off poverty.

Congress has been debating legislation to fix the private system, but it has been unable to resolve a basic tension: anything it does to ease the burden on failing or failed pension plans lessens the penalty for failure and enhances moral hazard. By making it easier for, say, a Delta or a Delphi to offer benefits, it raises the possible cost of a future bailout.

The tough medicine favored by the Bush administration, which would eliminate loopholes in the system as well as much of the subsidy that now exists in the insurance system, would lead to more companies freezing their plans or leaving the system outright. The number of pension plans would continue to shrink and in time all but disappear. This would strip the elderly of the future of what is still the most secure form of retirement income.

The fear of runaway pension costs plainly echoes the Social Security debate, and many suspect that the Bush administration would not much mind if pensions did disappear. "I don't think the administration is very interested in creating a future for traditional pensions," says Julia Coronado, a senior research associate at Watson Wyatt, a human-resources consulting firm. "It doesn't fit very well with their vision of the ownership society."

Bradley Belt, executive director of the P.B.G.C., shrugs off the charge. "The last thing we want to do is chase people out of the system," he says. Besides, the government doesn't need to chase. As Belt points out, the number of workers covered by pensions is shrinking without government help. In 1980, about 40 percent of the jobs in the private sector offered pensions; now only 20 percent do. The trend is probably irreversible, because it feeds on itself. Hewlett Packard, for instance, must compete with younger companies like Dell Computer that do not offer traditional pensions. Freezing its plan, which was a legacy of the company's famously employee-oriented founders, was an embarrassing step for H.P.'s present managers - but freeze it they did.

This may have made economic sense, but federal law has long recognized a social purpose to pensions as well. By allowing companies to deduct from taxes the money they contribute into their pension funds, the government encourages employers to provide a safety net for their workers. This remains a legitimate function, and if pensions were allowed to die, we would need something to take their place.

To understand why pensions are still important, you have to understand the awkward beast that benefits professionals refer to as the U.S. retirement system. It is not really one "system" but three, which complement each other in the crudest of fashions. The lowest tier is Social Security, which provides most Americans with a bare-bones living (the average payment is about $12,000 a year). The highest tier, available to the rich, is private savings. In between, for people who do not have a hedge-fund account and yet want to retire on more than mere subsistence, there are pensions and 401(k)'s. Currently, more than half of all families have at least one member who has qualified for a pension at some point in his or her career and thus will be eligible for a benefit. And among current retirees, pensions are the second-biggest source of income, trailing only Social Security.

During most of the 90's the decline in pension coverage was barely lamented. It was not that big companies were folding up their plans (for the most part, they were not) but that newer, smaller companies weren't offering them. As the small companies grew into big ones (think Dell, or Starbucks, or Home Depot), traditional pensions covered less of the private-sector landscape. This did not seem like a very big deal. Younger workers envisioned mobile careers for themselves and many did not want pension strings tying them to a single employer. And most were able to put money aside in 401(k)'s, often matched by an employer contribution.

It happened that 401(k)'s, which were authorized by a change in the tax code in 1978 and which began to blossom in the early 1980's, coincided with a great upswing in the stock market. It is possible that they helped to cause the upswing. In any case, Americans' experience with 401(k)'s in the first two decades of their existence was sufficiently rosy that few people shed tears over the slow demise of pension plans or were even aware of how significantly pensions and 401(k)'s differed. But 401(k)'s were intended to be a supplement to pensions, not a substitute.

From the beneficiary's standpoint, pensions mean unique security. The worker gets a guaranteed income, determined by the number of years of service and by his or her salary at retirement. And pensions don't run dry; workers (or their spouses) get them as long as they live. Because the employer is committed to paying a certain level of benefits, pensions are known as "defined benefit" plans. Since an individual's benefit rises with each year of service, the employer is supposed to sock money away, into a fund that it manages for all of its beneficiaries, every year. The point is that workers don't (or shouldn't) have to worry about how the benefit will get there; that's the employer's responsibility. Of course, the open-ended nature of the guarantee - the very feature that makes pensions so attractive to the individual - is precisely what has caused employers to rue the day they said yes. No profit-making enterprise can truly gauge its ability to meet such promises decades later.

A 401(k), on the other hand, promises nothing. It's merely a license to defer taxes - an individual savings plan. The employer might contribute some money, which is why 401(k)'s are known as "defined contribution" plans. Or it might not. Even if the company does contribute, it offers no assurance that the money will be enough to retire on, nor does it get involved with managing the account; that's up to the worker. These disadvantages were, in the 90's, somehow perceived (with the help of exuberant marketing pitches by mutual-fund firms) to be advantages: 401(k)'s let workers manage their own assets; they were a road map to economic freedom.

Post-bubble, the picture looks different. Various people have studied how investors perform in their 401 (k)'s. According to Alicia Munnell, a pension expert at Boston College and previously a White House economist, pension funds over the long haul earn slightly more than the average 40 (k) holder. Among the latter, those who do worse than average, of course, have no protection. Moreover, pensions typically annuitize - that is they convert a worker's retirement assets into an annual stipend. They impose a budget, based on actuarial probabilities. This might seem a trivial service (some pensioners might not even realize that it is a service). But if you would asked a 65-year-old man who lacked a pension but did have, say, $100,000 in savings, how much he could live on, he likely would not have the vaguest idea. The answer is $654 a month: this is the annuity that $100,000 would purchase in the private market. It is the amount (after deducting the annuity provider's costs and profit)  that the average person could live on so as to exhaust his savings at the very moment that he draws his final breath.

So the question arises: what if he lives longer than average? This is the beauty of a pension or of any collectivized savings pool. The pension plan can afford to support people who live to 90, because some of its members will expire at 66. It subsidizes its more robust members from the resources of those who die young. This is why a 401 (k) is not a true substitute. Jeffrey Brown, an associate finance professor at the university of Illinois at Urbana-Champaign and a staff member of the president's Social Security commission, notes that as baby boomers who have nest eggs in place of pensions begin to retire, they will be face with a daunting question: "How do I make this last a lifetime?  

The country's first large-scale pension plan was introduced after the Civil War, when the federal government gave pensions to disabled Union Army veterans and war widows. Congress passed an act in 1890 that extended pensions to all veterans 65 and over. This converted pensions into a form of social welfare. Over the next 20 years, states and cities added pensions for police officers and firefighters. By World War I,  most teachers had been granted pensions as well. Governments couldn't offer big paychecks for workers - teachers, the police, fighters - so it offered stability and pensions instead.

In the private sector, the first pension was offered by American Express, a stagecoach delivery service, in 1875. Railroads followed suit. Employees were required to work for 30 years before they qualified for benefits, and thus pensions helped companies retain employees as well as ease older workers into retirement. These employers thought of pensions as management tools, not as employee "benefts." But in the first half of the 20th century, as the historian James Wooten put it, government policies turned pensions into a tool of social policy. First came the tax deduction. This feature was abused, as abused, as companies used pensions to shelter payments to their executives. The rules were gradually tightened, however, forcing plans to include the rank and file. World War II gave more incentives to create pensions: punitive tax rates made the pension shelter enormously attractive and a government freeze on wages meant that pensions were the only avenue for increasing competition.

The effect of these policies was to encourage unions to bargain for pensions and to pressure employers to grant them. After the war, John L. Lewis, the legendary labor leader, staged a strike to win pensions for miners, Ford Motor capitulated to the U.A.W. in 1949. G.M., headed by the reluctant Sloan, followed in 1950. This led to a so-called pension stamped; by 1960, 40 percent of private-sector workers were covered. Meanwhile, in the auto industry, the seeds of the problem were already visible.

Companies might establish plans, but many were derelict when it came to funding them. When companies failed, the workers lost much of their promised benefit. The U.A.W. was acutely aware of the problem, because of the failing condition of several smaller car manufacturers, like Packard. The union didn't have the muscle to force full funding, and even if it did, it reckoned that if the weaker manufacturers were obliged to put more money into their pension funds, they would retaliate by cutting wages.

Thus in 1959, Studebaker, a manufacturer fallen on hard times, agreed to increase benefits - its third such increase in six years. In return, the U.A.W. let Studebaker stretch out its pension funding schedule. This bargain preserved the union's wages, as well as management's hopes for a profit, though it required each to pretend that Studebaker could afford a pension plan that was clearly beyond its means. Four years later, the company collapsed.

The Studebaker failure was a watershed. Thousands of employees, including some who had worked 40 years on the line, lost the bulk of their pensions. Stunned by the loss, which totaled $15 million, the U.A.W. changed its tactics and began to lobby in earnest for federal pension insurance. A union pension expert tellingly explained to Walter Reuther, the U.A.W. chief, that insurance would reconfigure the "incentives" of both labor and management. Though business was skeptical of the idea, a decade later, in 1974, Congress finally passed the Employee Retirement Income Security Act, or Erisa, which, among other protections, established the P.B.G.C. to insure private pensions. Erisa, according to Wooten, who wrote a history of the act, completed the transition of pensions into a part of the social safety net. It was also the birth of moral hazard.

Erisa, which would be amended several times, was supposed to ensure that corporate sponsors kept their plans funded. The act includes a Byzantine set of regulations that seemingly require companies to make timely contributions. As recently as 2000, most corporate plans were adequately funded, or at least appeared to be. Their assets took a serious hit, however, when the stock market tumbled. (In retrospect, they had been cavalier in assuming the bull market would continue.) And they were burned again when interest rates fell.

Since pension liabilities are, for the most part, future liabilities, companies calculate their present obligation by applying a discount rate to what they will owe in the future. As interest rates move lower, they have to set more money aside because it is assumed that their assets will grow more slowly. The principle is familiar to any individual saver: you need to save more if you expect, say, a 5 percent return on your investment instead of a 10 percent return. What is much in dispute is just which rate is proper for pension accounting.

Corporations have been gaming the system by using the highest rates allowable, which shrinks their reported liabilities, and thus their funding requirements. The P.B.G.C., when calculating the system's deficit, uses what is in effect a market rate; whatever it would cost to buy annuities for everyone covered in a pension plan is, it argues, the plan's true "liability." The difference between these measures can be extreme. Depending on whom you talk to, General Motors' mammoth pension fund is either fully funded or, as the P.B.G.C. maintains, it is $31 billion in the hole.

What is not in dispute is that when interest rates fell, the present value of pension liabilities (by whatever measure) soared. The confluence of falling stock prices, plunging interest rates and a recession in the beginning of this decade was the pension world's equivalent of the perfect storm. An unprecedented wave of pension sponsors failed and then dumped their obligations on the P.B.G.C. (To do so, a sponsor generally must prove that it could not re-emerge as a viable enterprise without shedding its pension plan.) By far the most costly failures were in airlines and steel, although the list ranges from Kemper Insurance and Kaiser Aluminum to Murray, a lawn-mower manufacturer.

As the P.B.G.C. assumed responsibility for more and more pensioners, it became clear that the premium it charged was way too cheap. Mispriced insurance, like mispriced anything, sends the market a distorted signal. Belt, the P.B.G.C. director, who served as counsel to the Senate Banking Committee in the late 1980's during the savings-and-loan crisis, says that cheap pension insurance gave rise to flawed incentives: namely it kept companies in the pension business who didn't deserve to be there. He also argues, rather convincingly, that lax rules allowed pension sponsors to get away with inadequate funding.

For example, United Airlines did not make contributions to any of its four employee plans between 2000 and 2002, when it was heading into Chapter 11, and made minimal contributions in 2003. Even more surprisingly, in 2002, after two of its jets had been turned into weapons in the Sept. 11 disaster, and when the airline industry was pleading for emergency relief from Congress, United granted a 40 percent increase in pension benefits for its 23,000 ground employees.

Bethlehem Steel similarly enjoyed a three-year funding holiday as it was going through hard times, letting its liabilities swell in advance of turning them over to the government. Meanwhile, in order to gain its unions' approval for plant shutdowns, it agreed to costly benefit enhancements. In 2001 Bethlehem filed for Chapter 11 bankruptcy. It was guided through its bankruptcy by none other than Miller, now the Delphi C.E.O. Miller disputes the notion that capital-scarce companies like Bethlehem intentionally game the system by shirking funding. "Companies don't like falling behind," he says. "When you have a hard choice between starving the capital base to feed the pension plan, or making capital investments to become more productive, to the extent there is permission that's what you do." The point is, they had permission.

Neither Bethlehem nor United broke any laws. Both companies made the full contributions required under Erisa. When the P.B.G.C. seized their plans, however, Bethlehem was only 45 percent funded, and United was only 42 percent funded. For companies that terminate their pension plans, such gross underfunding has become the norm. Either assets suddenly vanish when the P.B.G.C. walks in the door, or, evidently, the system for measuring "full" funding is broken. As Belt testified to the Senate Committee on Finance in June, "United, US Airways, Bethlehem Steel, LTV and National Steel would not have presented claims in excess of $1 billion each - and with funded ratios of less than 50 percent - if the rules worked."

Even leaving aside the debate over which rate to use in calculating pension liabilities, there is no doubt that Erisa permits companies to use some doubtful arithmetic. For instance, the law lets corporations "smooth" changes in their asset values. If the stocks and bonds in their pension funds take a hit (as happened to just about every fund recently), they don't have to fully report the impact. Nor do they have to ante up fresh funds to compensate for the loss for five years. A similar smoothing is permitted on the liability side. And though, in theory, Erisa discourages underfunding by requiring offenders to pay higher premiums, its various loopholes render the sanction toothless. Thanks to another loophole, companies that contribute more than the required amount get to skip future contributions even if they later become underfunded. These companies are awarded so-called "credit balances," which remain in place even if the actual balance is showing red.

Incredibly, when United's plans were terminated, earlier this year, even though they were groaning under $17 billion in pension liabilities and a mere $7 billion in assets, they still had credit "balances" according to Erisa. (By law, the P.B.G.C. will be on the hook for most, but not all, of United's shortfall. The agency guarantees pensions up to $45,000 a year; employees, mostly pilots, who were owed richer pensions are uninsured above the cap.)

Their dubious funding history notwithstanding, corporations - airlines in particular - have been lobbying for greater permissiveness for several years. And they have gotten it. Congress has twice relaxed the rules, permitting pension sponsors to use a higher rate to calculate their liabilities.

Enter the Bush administration: it has essentially declared the era of permissiveness over. Among other changes, it wants the funding rules tightened. To tackle moral hazard, it wants to stop companies with poor credit ratings from granting benefit hikes, or from doling out unfunded pension benefits to unions who agree to plant shutdowns. It even wants to prevent workers at some companies whose bonds are given a "junk" rating from accruing more years of service. This would be painful to employees at many industrial companies, possibly including G.M.

Indeed, one reading of the administration proposal is that, having seen the steel and airline industries raid the P.B.G.C., it is drawing the line at the auto industry - whose initial distress, of course, prompted the agency's founding. Asked about that before Delphi went bust, Belt admitted: "Eight auto-parts suppliers have come under Chapter 11 so far this year. No question our single largest source of exposure is the auto sector."

Since G.M.'s stock was downgraded to junk status earlier this year, the possibility that it would file for bankruptcy has been the subject of on-again, off-again debate on Wall Street. G.M.'s pension plan totals an astronomical $90 billion; a bankruptcy filing would be the P.B.G.C.'s biggest nightmare. G.M. says the notion is far-fetched. The company seems to have plenty of liquidity and, just two weeks ago, with retiree costs a major concern, it reached an agreement with the U.A.W. to trim health benefits. G.M. and other industrial companies, along with their unions, have harshly attacked the Bush pension proposal, which would force many old-economy-type corporations to put more money into their pension funds just when their basic businesses are hurting.

Alan Reuther, Walter's nephew and the U.A.W.'s legislative director, says the provisions to restrict benefits would be "totally devastating for workers and retirees." He makes no apologies for "30 and out" - a fair reward, he maintains, for hard service on the assembly line - and he wonders at the post-modern notion that blue-collar workers should be responsible for their own retirements because giant corporations can't handle it. Also, a typical G.M. pension for someone with 30 years on the job is about $18,000 a year. That is hardly to be compared with an airline pilot's. "The P.B.G.C. is focused on protecting themselves from claims and not on protecting the claims of workers," he says. "They forget why they were created." Social safety nets have their price - in this case, a little moral hazard - and that is really what the debate is about.

What has emerged from the Beltway skirmishing thus far are bills on either side of Congress that would in some ways tighten funding but give a special break to airlines. Premiums to the P.B.G.C. would rise from $19 per plan participant to $30, and variable premiums on distressed companies would be enforced. The bills would chip away (but not eliminate) gimmicks like "smoothing."

The Senate is still divided, however, on how to treat corporations with junk credit ratings - the ones most likely to wind up in the P.B.G.C.'s lap. Hard-liners like Senator Chuck Grassley insist they should be forced to strengthen their pension plans in a hurry; Senators Mike DeWine and Barbara Mikulski (both from states with blue-collar constituencies) want to give such companies lenience. So after months of lobbying, politicking and deal making, moral hazard is still alive.

The P.B.G.C. does not protect government pensions, but dynamics similar to those in the private sector have also wrecked the solvency of public plans. Even in states where budget restraint is gospel, public-service employees have found it relatively easy to get benefit hikes for the simple reason that no one else pays much attention to them. In the corporate world, stockholders, at least in theory, exert some pressure on managers to show restraint. But who are the public sector "stockholders"? The average voter doesn't take notice when the legislature debates the benefits levels of firemen, teachers and the like. On the other hand, public-employee unions exhibit a very keen interest, and legislators know it. So benefits keep rising.

As a matter of practice, those benefits are as good as insured. Because public pension benefits are legally inviolable, default is not an option. Sooner or later, taxpayers will be required to put up the money (or governments will be forced to borrow the money and tax a later generation to pay the interest). Thus, unions can bargain for virtually any level of benefits without regard to the state's ability, or its willingness, to fund them. This creates moral hazard indeed. At least in the private sphere, there are rules - ineffectual rules maybe, but rules - that require companies to fund. In the public sector, legislatures wary of raising taxes to pay for the benefits that they legislate can simply pass the buck to the future. This explains how the West Virginia Teachers Retirement System has, embarrassingly, only 22 percent of the assets needed to meet its expected liabilities. It also explains how Illinois, a low-tax state, is underfunded by some $38 billion, or $3,000 per every man, woman and child in the state.

California is a good example of the political forces that have driven benefits higher. In the 90's, Gov. Gray Davis, a Democrat who was strongly supported by public-employee unions, pushed through numerous bills to increase benefits. One raised the pension of state troopers retiring at age 50 to 3 percent of final salary times the number of years served. (Previously, the formula was 2 percent at age 50, more if you were older.) Thus, a cop hired at age 20 could retire at 50, find another job and get a pension equal to 90 percent of his final salary.

The higher benefits trickled down to the local level, as counties that feared losing police officers to the state felt forced to copy the formula. Counterintuitively, as benefits were going up, the California Public Employees Retirement System (Calpers), which was boasting high returns in the stock market, allowed state agencies and local governments to reduce their contributions.

Contra Costa County, which adopted the "3 percent at 50" formula for its Police Department, got by with contributing only $55 million to retirement costs in 1999, near the market peak. When the market tanked, the county found itself with lower assets and greater obligations. Six years later, the county's retirement bill had more than tripled to $180 million. Bill Pollacek, the county treasurer-tax collector, says the excess earnings from the bull market were spent, among other things, on higher benefits; "the losses were left for the taxpayers."

This example was repeated with various twists across the country. In New Jersey, for example, Christine Whitman, the Republican governor in the 90's, ultimately relied on buoyant stock-market predictions to finance hefty tax cuts, which were the centerpiece of her administration. In 1997, New Jersey borrowed $2.8 billion, at an interest rate of 7.64 percent. The money was advanced to its pension system, on the convenient theory that its pension managers would make more in the market than the state paid out in interest. For a while, they did. The state even raised benefits.

Meanwhile, Trenton achieved a sort of transitory budget balance by contributing less to its pension system. New Jersey's contribution to the Police and Firemen's Retirement System was zero in 2001 through 2003. But during the dot-com debacle, its investments plunged. And the state came under intense budget pressure because of the recession, and so gave itself a few years more to start paying down its pension liability (which further widened the gap). This year, the last easy-funding year, New Jersey will contribute $220 million to its pension system; by 2010, the annual bill will be an impossible-seeming $2.5 billion.

I spoke to Jon Corzine and Doug Forrester, the candidates in next Tuesday's gubernatorial election, and while each expressed the proper horror with regard to past mismanagement, neither had much to say about how they would replenish New Jersey's pension system. State pension officials say that if New Jersey were a private corporation, its system would be nearly bankrupt. "In the real world this is a P.B.G.C. takeover," Fred Beaver, the head of the pension division, told me. Raising taxes is politically forbidden (Forrester has been campaigning to cut property-tax rates).

And the state's reported pension underfunding, officially $25 billion, is undoubtedly optimistic. It assumes that New Jersey's pension assets will earn 8.25 percent, a number collectively determined - some say pulled from thin air - by the state's pension council. Even Orin Kramer, a private hedge-fund manager who also is also chairman of the council, says that any assumption higher than 7.5 percent is unrealistic. "The published numbers are divorced from economic reality," Kramer says. "No one even does the math for what will happen if you only do 7 percent because it's too serious. You start firing cops and teachers."

According to Barclay's Global Investors, if you use realistic assumptions, the total underfunding in all public plans is on the order of $460 billion. If this figure is even close to true, future taxpayers will be hopelessly in hock to the police, firefighters and teachers of the past.

Cutting pensions (unlike health benefits) is simply not an option. State constitutions forbid public entities, even prospectively, from reducing the rate at which employees accrue benefits. They can tinker with, or abolish, benefits for future employees, as Alaska did, but for a worker already on the payroll, benefits - even benefits that might not be earned for many decades hence - are sacrosanct. These benefits are like headless nails; once driven in they can never be removed. This year, New York's Legislature approved 46 new bills - more headless nails - to increase pension benefits, according to E.J. McMahon, an analyst at the Manhattan Institute. New York's benefits already rank among the most generous in the country, and the new bills would expand categories of workers who can retire early, or who can qualify for higher rates. Such bracket creep is pervasive.

ONE OF THE BIGGEST pension offenders is San Diego, where six members of the pension board, including the head of the local firefighters' union and two other union officials, have been charged with violating the state's conflict-of-interest code, a felony. What is interesting about San Diego is that, juicy details aside, its pension mess actually looks rather commonplace. The six board members are accused of making a deal to let City Hall underfund the pension system in return for agreeing to higher benefits - including special benefits for themselves. Explicitly or otherwise, this is what unions and legislators have been doing all over the country. A senior adviser on pensions to Gov. Arnold Schwarzenegger told me he fears that ever higher benefits are inescapable, given the fact that legislators control the benefits of people whose support is vital in elections.

Calpers, the country's biggest state-employee retirement system, responds that the pension system has worked well. And for Calpers's 1.4 million members, it has. The average benefit for retirees is $21,000 a year, more than most at General Motors. But at some point, the interest of the public and the interests of public employees diverge.

Earlier this year, Schwarzenegger tried to move California to a 401(k)-style defined contribution plan (for new employees), but the Legislature refused to go along. Schwarzenegger has vowed to revisit the issue in 2006. This battle is being fought from statehouse to statehouse. Michigan (mimicking Alaska) has closed its pension plan to some new employees, and various states, including Florida, Colorado, South Carolina, Arizona, Ohio and Montana, are taking a partial step of letting employees choose between defined contribution plans and traditional pensions. This compromise does not really change much. Most employees who are given the choice opt, quite naturally, to keep their pensions.

Partly for that reason, the Citizens Budget Commission, a politically neutral watchdog, concluded that only by ending pensions outright (for new employees) could New York avert a future fiscal calamity. "Changes in pension benefits for future workers would yield fiscal gains only slowly," the commission noted in a position paper, "but the service to the future fiscal health of the City and State would be enormous."

Most legislatures are not about to do that anytime soon. There is a legitimate argument for preserving public pensions, however, if only they could be put on a sound fiscal basis. Critics like Grover Norquist, the tax-cut crusader, lampoons pensions as remnants of a stodgy, Old World economy. The desire to collect a pension, he argues, keeps workers from moving to better opportunities and shackles employers to workers who are just marking time.

But while mobility is generally considered a virtue in the modern economy, it isn't appropriate everywhere. It may be desirable for a software engineer to move from job to job, notes Robert Walton, a Calpers assistant executive; "for teachers, firefighters, nurses, engineers, that isn't the type of work force you want." Stability is a virtue. The trick is to force legislatures to commit to funding with the same zeal with which they commit to benefits.

De Maio, the San Diego watchdog, is lobbying for a federal law that would impose Erisa-type rules on public plans. Another solution might be found in the Texas Municipal Retirement System, which represents 800 cities and towns in the state. It has a blended system of automatic employer and employee contributions that are managed by the system and turned into an annuity upon retirement. These sorts of remedies could avert plenty of future San Diegos. In principle they are quite simple. It is only the politics that are difficult.

On the private side, benefits professionals have been touting so-called cash-balance plans, a hybrid that in some ways looks like a 401(k), as the best hope for saving the pension industry. With a traditional pension, employees accrue benefits very slowly during their first 20 years and very rapidly during their next 10 (this is why pension plans act as retention tools; you pay a penalty for leaving early). Thus, an employee who stays at a company for 30 years gets a much bigger pension than one who works at three companies for 10 years each. Cash-balance plans were devised to appeal to younger workers, most of whom do not envision retiring at the firm that hired them out of college. In these plans, employees accrue benefits steadily, one decade to the next. There is no penalty for leaving, and workers who change jobs simply roll their accrued benefits into their next plan, as with a 401(k). Many firms converted to cash-balance plans in the 90's to attract younger and more mobile workers.

But the downside of giving more to junior employees is that senior employees get less. When I.B.M. converted, it reduced the rate at which some employees of long standing would accrue benefits, touching off a firestorm. The company was sued, I.B.M. lost and the legal status of similar plans remains in doubt. The pension industry has been lobbying Congress to clarify the status of existing cash-balance plans, but neither the administration nor anyone on Capitol Hill has done so.

To some people, this is further evidence that the Bush administration would just as soon be done with pension plans altogether. I put that recently to Elaine Chao, the secretary of labor, and while her answer was diplomatic, she made no bones about the fact that, in the administration's view, traditional pensions are losing their relevance. "Defined benefit plans have their advantages," she told me, "but in an increasingly mobile 21st-century work force, the lack of flexibility of D.B. plans is yielding to greater usage of defined contributions plans."

It's hard to argue with her, if you look at the numbers. Although 44 million people are covered by private-sector plans, half are people who have already retired and are collecting benefits or whose plans have been frozen or terminated. In other words, on-the-job employees accruing benefits - once the backbone of the system - constitute only half. At that rate, even without legislation, the private-sector pension community will mostly die off in a generation.

And pension sponsors are likely to get another jolt soon. Under current accounting standards, companies can "smooth" their earnings reports, so that each quarter's net income reflects the average assumed performance of the company's pension assets, whether up or down, but not the actual performance. (Discrepancies from the average are sifted back into the earnings stream over time.) This means that reported earnings are often wildly misleading. Robert Herz, chairman of the Financial Accounting Standards Board, has criticized this practice as "a Rube Goldberg device." If FASB follows up and disallows it, corporate pension sponsors would have to cope with a lot more volatility in their earnings. Managers hate volatility, and such a change would prompt many of them to fold their plans.

If defined benefits are on their last legs, then it would make sense to try to incorporate their best features into 401(k)'s. The drawback to 401(k)'s, remember, is that people are imperfect savers. They don't save enough, they don't invest wisely what they do save and they don't know what to do with their money once they are free to withdraw it. Quite often, they spend it.

Here there is much the government could do. For instance, it could require that a portion of 401(k) accounts be set aside in a lifelong annuity, with all the security of a pension. Behavioral economists like Richard Thaler have demonstrated that you can change people's behavior even without mandatory rules. For instance, by making a high contribution rate the "default option" for employees, they would tend to deduct (and save) more from their paychecks. If you make an annuity a prominent choice, more people will convert their accounts into annuities.

Otherwise, it's not hard to predict that as octogenarians and nonagenarians become commonplace in society, many are going to outlive their savings, which is even more scary than outliving the savings of the P.B.G.C. Promoting an annuity culture is probably the single best way to make up for the demise of pensions. Yet most companies that provide 401(k)'s don't even give the option of purchasing an annuity when people cash in their accounts. As Brown, the Illinois professor, notes, "There is no box to check that says 'annuities."' That is a minor scandal. "I wish someone in Washington were thinking bigger thoughts about what the optimal retirement package should look like," says Watson Wyatt's Coronado.

What are Secretary Chao's thoughts? She bounced the question to the Treasury Department. Mark Warshawsky, the Treasury's top economist, has written about the need for annuities, and in an interview he allowed that as 401(k)'s become the primary, or the only, source of retirement income for more people, "I think it is a concern that annuities are not being offered in those plans." When I asked what the Treasury was doing about encouraging annuities, Warshawsky merely said that it was under study. Anything that smacks of regulation (like rules to make sure employees get a particular menu of choices, whether for annuities or for their portfolios) gives the administration shivers. This is what you would expect, given the administration's strong free-market tendencies.

But the government is already deeply involved, since it shelters retirement savings - pensions, yes, but also 401(k)'s, which are similarly permitted to grow tax-free. When it passed Erisa, Congress agreed that corporations that invested tax-sheltered retirement funds - pensions - should have to live by certain rules. But in the defined contribution world - the world of 401(k)'s - there are no rules. Employers can contribute or not. Employees can diversify or blow it all on the company stock (even if it is Enron). If nothing else, the century-long experiment with pensions has proved that in the absence of the right rules, the money will not always be there. The purpose of pension reform should be not merely to avoid a fiscal disaster but to find a fiscally sound way to preserve the likelihood of secure retirements. If people are going to retire on 401(k)'s, those should be subject to rules, and guidance, as well.

It would be nice to think that reform would include a future for pensions, but on the private side at least, it is doubtful. As Delphi's Miller put it simply: "A pension plan makes no sense in today's world. It's not wise for a company to make financial promises 40 or 50 years down the road." Most American executives would agree. Miller says he has not decided what to do at Delphi. If workers grant wage concessions, he has said, the pension plan, which is $4.5 billion shy of what it needs, might even survive. This has the sound of a bargaining ploy. Knowing that the P.B.G.C.'s guarantee is in place, the unions will probably insist on keeping their wages as close to intact as they can, and Miller will probably end up handing the pension plan over to the agency, just as he did at Bethlehem. Then, Miller and other executives will get stock and dandy bonuses in a new Delphi that is happily stripped of pension obligations, and some 45,000 employees and retirees will, in time, happily collect their pensions - courtesy of the U.S. Government. Moral hazard at work.

Roger Lowenstein, a contributing writer, has written about Social Security and health care reform for the magazine.

Copyright 2005 The New York Times Company. Reprinted from The New York Times Magazine of Sunday, October 30, 2006.