ONE BIG reason that it's so hard to get a firm hold on the federal budget is that a very large share of it is tied to increases in the cost of living. Over the last decade, inflation-linked benefits have forced other programs to compete for a shrinking budget share. The squeeze has now become so tight that the administration and Congress can no longer avoid the difficult question--more political than technical--of how much insurance the society should provide against loss of purchasing power.

Inflation-linked benefit programs now account for close to two-thirds of the non-defense budget. This includes both programs like Social Security that are automatically tied to the Consumer Price Index and others--like Medicaid, Medicare and low-income housing--whose costs are driven by price increases in particular sectors.

Nobel laureate James Tobin remarked the other day that Congress did not act in total ignorance of what it was doing when it created these benefit entitlements. It meant--at least in some vague way--to make sure that certain groups did not suffer unfairly from inflation. But Congress certainly didn't think through all the consequences of what it was doing. As a result, inflation-indexed programs have grown much faster than anyone expected and there have been results that no one probably intended.

Some unexpected results came from simple mistakes. The formulas for computing initial Social Security and federal retirement benefits had technical flaws, and the inflation measure used to index benefit payments--the standard CPI--tended to exaggerate average housing costs. These technicalities have now been corrected, although their costs have not been recouped.

But choosing the proper index for a program is not primarily a technical problem. The "right" choice depends on political judgments. What do you want to insure: the ability of people to buy exactly what they are buying now? Protection of a minimally adequate standard of living? Guaranteed access to a reasonable level of medical or housing services? Whom do you want to protect: the aged? The poor? Home buyers? Under what conditions: continued economic growth? General economic decline?

No one, for example, thought about what would happen to indexed programs during a time--such as the last decade--when the average wage earner was losing ground to inflation. As Prof. Tobin noted, you can't protect everyone in the country against the loss of real income caused by an OPEC oil shock. When the shocks came, some people--those with inflation-adjusted incomes or strong market power--were able to avoid a loss of purchasing power. Everyone else then had to absorb not only his own losses but the losses that properly belonged to the favored few.

At the time that Social Security was indexed, no one asked whether, during prolonged bad times, the retired should gain at the expense of the average wage earner. If the question had been debated openly, it is fair to say that a somewhat different answer might have been reached. That's one of the stickiest but most important issues that need to be pursued during the next round of budget cuts.