It is the world's largest money pump, one of the main engines in the U.S. economy.

More than 100 million people pay into it each year.

More than 36 million, a seventh of the population, depend on it for all or part of their income.

In the last 20 years it has helped halve the percentage of older Americans living in poverty.

But its costs have multiplied in the process; it now makes up more than one-fourth of the federal budget, is larger than all but about a dozen economies in the world and is in trouble.

This largest and in many ways least understood of all federal programs is Social Security. It consists of three great trust funds, each with a stipulated share of the Social Security tax:

* An old-age and survivors fund, by far the largest, which will pay out $155 billion next year.

* A disability fund, which will pay out $19 billion.

Medicare, which will expend about $55 billion to help pay hospital and doctor bills of the aged and disabled.

The disability and Medicare programs are healthy now, though Medicare may start to need shoring-up in the 1990s.

The problem is in the old-age fund. It cannot keep borrowing from the other two and now seems likely, absent action by the president and Congress, to run out of money in 1984.

At one level the choices the president and Congress have are simple: raise the tax, already 6.7 percent for both employer and employe on the first $32,400 of every covered worker's wages, or curtail projected costs by cutting benefits.

At another level these choices are complex. They may be an example of the intergenerational politics and politics of scarcity that some see ahead as the population ages and economic growth tails off. They also represent a painful set of decisions about the distribution of income, including how much, and how, the country ought to save and invest instead of spend.

The drying-up of the old-age fund has occurred because of the failure of an assumption.

In 1972 Congress put Social Security on automatic pilot, indexing both taxes and benefits; that is, tying them to inflation so they would rise automatically in presumed lock step in the future.

Taxes were tied to wages, benefits to prices, and it was assumed that wages and prices would continue to behave as they had in the past. That is, wages would tend to rise about 1 1/2 percentage points more than prices each year. Wages thus would provide an ever-bountiful base to tax and let benefits rise.

The inflation rate in that election year seemed to bear out this optimism. It was 3.4 percent. But the next year the government's price indices broke the mold. Food prices rose 20 percent in 1973, and consumer prices almost 9 percent. In the winter of 1973-74 the Organization of Petroleum Exporting Countries knit itself together, slapped an embargo on shipments to the United States in retaliation for U.S. support of Israel, then tripled the world crude oil price.

The U.S. inflation rate in 1974 was 12.4 percent, with wages lagging far behind.

Prices have continued to outstrip wages in most recent years. Unexpectedly high unemployment rates have also cut into Social Security revenues. The lock step hoped for in 1972 has not occurred.

Jimmy Carter tried to deal with this and reorder the Social Security system in 1977, mainly through an enormous tax increase. Even that has not proved enough.

Last year President Reagan also turned to the problem, proposing not a tax increase but large cuts in benefits. His recommendations created an uproar.

The precarious financial condition of the old-age fund is one of the most volatile issues in domestic politics. More than two-thirds of the old people in the country now look to Social Security for half or more of their income, and if Social Security does not support them their children will have to.

Democrats charged Reagan's proposed cuts were far larger than needed to shore up Social Security, and that he was trying to balance the budget on the backs of the elderly.

Reagan, who has been periodically wounded by Social Security proposals throughout his political career, ultimately withdrew the benefit recommendations and named a bipartisan commission -- 10 Republicans and five Democrats -- to study the system.

The commission, whose chairman is respected Republican economist Alan Greenspan and chief of staff is Robert J. Myers, former Social Security chief actuary, is expected to make its recommendations in mid-November, just after the Nov. 2 election.

The panel's membership suggests it will propose gradual and cautious rather than drastic changes. "I don't see any bomb-throwers sitting around this table," said one commission member, surveying Greenspan, Senate Finance Committee Chairman Robert J. Dole (R-Kan.), Aging Committee Chairman John Heinz (R-Pa.), and ranking House Ways and Means Committee Republican Barber B. Conable Jr. (N.Y.).

"No one I know of in either party wants to do anything but preserve Social Security," Dole said recently.

Thus, one likely proposal is to advance some or all of the Social Security tax increases scheduled over the next several years under the Carter tax bill in 1977.

These would lift the tax rate to 7.65 percent for both employer and employe by 1990.

The wage base -- the amount of each individual's earnings to which the tax applies -- is also scheduled to keep rising each year.

Alice Rivlin, director of the Congressional Budget Office, has said Social Security could raise an extra $46 billion in 1984 and 1985 by making all these tax increases effective Jan. 1, 1984.

Some Democrats have said that, instead of increasing the Social Security tax, they would rather start using the income tax to help defray Social Security costs. Their reason is that the income tax is progressive -- rates rise with income -- while the Social Security tax is flat.

Conservatives, however, have resisted using general revenues for Social Security, saying it will remove one of the few disciplines left on the program. Rather than raise taxes, they would prefer, as Reagan did, to work on benefits.

One proposal is to alter the indexing formula and limit future cost-of-living increases. One idea by Myers is to legislate the level of increase that the theoreticians assumed would occur in 1972, or let benefits rise each year 1.5 percent less than wages. Among other things, this would eliminate the Social Security inequity of the last few years, in which people receiving the benefits have stayed even with inflation while wage-earners paying the tax have fallen behind.

Still other alternatives include:

* Bringing state and local government employes into the program. They would all start paying taxes now, but most would not start collecting benefits until later. The result would be a big cash infusion for the system. Congress this year took a similar step and voted to make federal employes start paying the Medicare part of the Social Security tax.

* Gradually shifting the normal retirement age from 65, where it is now, to 67 or 68 after the turn of the century. But this would not help the system much right away. This might also involve gradually increasing the benefit reduction that comes with early retirement. Reagan proposed last year that the benefits of early retirees be quickly and rather drastically cut below present levels. This was one of the most controversial of his proposals.

* Gradually reducing the so-called replacement ratios of Social Security out into the future, the amount of a worker's last paycheck replaced by his first benefit check. In effect, this would amount to an across-the-board cut in future basic benefit levels.

The Social Security benefit structure is progressive, in that the replacement ratio is higher for low-income workers than for those with higher wage histories. The replacement ratio for the lowest-income workers now is a little over 50 percent, for the average worker a little over 40 percent and for a worker who earned at the tax cutoff all his career about 27.5 percent.

The average benefit now for a retired worker is $416 a month, the maximum, $729 or $8,748 a year.

* Taking steps to curb future Medicare costs. The Medicare program, while in good shape now, is expected also to run into trouble in the 1990s. There are numerous proposals for reducing future outlays, ranging from simple limits on how much the government will pay for certain services to the fostering of more competition in the health care field, which some economists think would cut into medical inflation.

Conservatives also are resisting proposals to finance Medicare out of general revenues.

One question is whether the commission, in making its recommendations, will confine itself to Social Security's so-called short-term problem -- the one caused by higher-than-anticipated inflation and unemployment in recent years -- or will also turn to the long-term problem, which has to do with demographics.

This problem is familiar. The baby boom generation will reach retirement age starting around the year 2020.

The lower birth rates of the post-baby boom era are leaving a smaller generation of workers to pay the taxes needed to support the large numbers of retirees. Today there are 31 beneficiaries per 100 active workers who are paying payroll taxes into the system. The change in population mix should force it to rise to 50 beneficiaries per 100 workers.

This demographic change is why the old-age and disability trust funds are projected to start declining rapidly after 2020 and go broke by 2030.

The cost of old-age and disability benefits now is equal to about 11.8 percent of the total payroll in the nation subject to the Social Security payroll tax. By 2060, the cost will be 16.81 percent of taxable payroll each year, but the payroll tax now scheduled to cover the costs of these two programs will be only 12.4 percent (6.2 percent each for employers and employes.)

Under the pessimistic scenario, the cost of old-age and disability benefits in 2060 is estimated at 28.5 percent, which means workers and their employers would each be paying a payroll tax of more than 14 percent to maintain the programs, and this does not include the cost of Medicare.

It should be noted that many Democrats, including Robert Ball, former Social Security commissioner (1962 to 1973) and a member of the commission, do not accept all the long-term demographic and economic assumptions.

Ball has repeatedly told the commission that demographic and economic projections much beyond 25 to 50 years are, in effect, like reading tea leaves and should not be used as a reason to make long-term cuts in the system.

Moreover, Ball argues that while there will be fewer workers to support each retiree, the number of children per family will be fewer than now so a larger portion of an individual's resources would be free to help support Social Security.

And he argues that while financing of the system may lag, the same projections show that as percentage of the gross national product, the cost of the old-age and disability programs, now 5.16 percent, will barely rise, to only 5.44 percent in 2060. This means that society could pay for the costs out of its greater productivity.