The president, to his credit, has ignited a roaring debate about Social Security reform – the proverbial third rail of politics. Most would agree that Social Security faces a long-term funding crisis spurred by the aging of the population combined with large scheduled increases in benefits. The program is expected to spend more than $500 billion this year, accounting for more than 20% of the federal budget and 4.2% of gross domestic product. By 2050, an unchanged system is projected to reach 6.5% of GDP while Medicare would add almost another 10% of GDP.

Social Security has been in trouble before. With hindsight, the policies adopted to resolve past crises were politically expedient, and at best did nothing and at worst aggravated the long-term situation. We are more likely to find a better solution this time if we learn from past mistakes. Although the program is not expected to run cash deficits until 2017, changes to the system must be scheduled soon. Plans affecting future retirement income should be spelled out well in advance, allowing today’s young workers to adjust their lifetime work and savings patterns.

How We Got Where We Are

From the start, Social Security had a muddled mission. Early on, President Roosevelt emphasized the prevention of poverty in old age as the major reason for an “old age insurance system” for workers. But the program soon moved well beyond Roosevelt’s initial anti-poverty objective. As Wilbur Cohen, a major architect of the program put it, “A program that is only for the poor – one that has nothing in it for those with middle and upper incomes – is in the long run a program the public won’t support.” Over the years, average incomes grew rapidly and Social Security benefits grew even more rapidly. Today, only a minor portion of benefits goes to eliminating poverty.

No major fiscal crisis occurred until the 1970s. Benefits were not indexed and legislated increases in benefits were matched by increases in the payroll tax. Benefits were relatively low. The replacement rate – the initial monthly benefit awarded to new retirees as a percent of their monthly earnings before retirement, was much lower than it is today.

A sea change occurred in the 1970s. Benefits and replacement rates increased sharply, spurred by Wilbur Mills, the powerful chairman of the House Ways and Means Committee, who tossed his hat in the ring for president in 1972, promising: “If you want a 20 percent increase in Social Security benefits, vote for Wilbur Mills.” And he kept his word. (Mills’s political ambitions ended in 1974, after his involvement with a stripper was disclosed when she jumped from Mills’s limo and ran into the Tidal Basin.)

The 1972 legislation pushed by Mills also mandated automatic annual indexing of benefits for inflation – both for current retirees (cost of living adjustments) and for initial awards of new retirees. Unfortunately, a flaw in the method used to adjust initial awards produced over-indexing for inflation and triggered an explosive increase in the benefits of new retirees.

The debate in the 1970s over how to fix the over-indexing flaw considered two methods: price-indexing and wage-indexing. Price-indexing would have adjusted initial awards only for inflation. As average real earnings increased over time, average real benefits would also increase, but not by as much as earnings, because the underlying benefit formula is progressive, much like our income tax system. Wage-indexing by contrast, overrides the progressivity in the benefit formula by adjusting initial benefits for both inflation and real wage growth. Consequently, the benefit awarded to the average new retiree rises in step with economy wide earnings, no matter how high earnings get.

Wage-indexing with its higher benefits won the day. Of course, higher benefits meant higher costs. But the specter of rising costs was disregarded and wage indexing became law in 1977.

The escalation of benefits in the 1970s combined with a sluggish economy spawned the next fiscal crisis. Benefits exceeded tax receipts starting in 1975, and continued to do so through 1983. The bipartisan commission appointed to resolve the crisis (headed by Alan Greenspan) made many recommendations that became law in 1983. However, the single major contribution to improving the long-term outlook was to increase the age of retirement gradually from 65 to 67, starting in 2003. In addition, the payroll tax was raised and Social Security benefits were made partially subject to the income tax.

The infusion of taxes assured that tax receipts would exceed benefits, producing surpluses up to about 2017. But despite popular belief, those surpluses cannot fund the cash deficits that will grow sharply after 2017, because they already have been used to fund non-Social Security spending and will continue to do so until 2017. Our pay-as-you-go system has no good mechanism for prefunding benefits.

Where Do We Go From Here?

If we were starting all over, the ideal government retirement system would be a safety net to prevent poverty among the elderly, the most pressing reason for a federal program. Another reason often given is that people are myopic, and if left to their own devices would not accumulate enough assets for their old age. But if the public wants a government system of mandatory savings, a prefunded system in the form of individual accounts would be the only sensible choice. If a prefunded system had been in place when the baby boomers began working, their benefits now would be fully funded by the real assets accumulated in their private accounts. There would be no financial crisis.

However, we are not starting from scratch and the ideal is not likely to be politically or practically feasible. A few limited goals must be achieved. One is to contain the system’s costs. The most straightforward way to do this is to switch to price-indexing for adjusting the initial benefits of future retirees. If price indexing were the only change in the system, the long-term costs of the program would grow more slowly, and, depending on the particular indexing method, would require small – if any – increases in tax rates. Lower benefits and taxes would give workers both the incentive and wherewithal to invest in private plans to enhance their retirement income and would also encourage increased work participation.

Several proposals have recommended structural changes that would combine a price-indexed (but smaller) pay-as-you-go system with a prefunded retirement savings program by diverting a portion of payroll taxes into private accounts. Society gains from individual accounts because prefunded benefits reduce future taxes, and the real savings generated boost economic growth. True, those gains have a cost – the “transition cost” that arises when a portion of taxes is diverted to private accounts and additional funds must cover a portion of current retiree benefits. Those costs should be viewed as an investment. By contrast, the costs of the current system are permanent, not transitional, and will only grow more intractable with time.

Ms. O’Neill is a professor of economics at Baruch College of the City University of New York and a former director of the Congressional Budget Office.