THE FEDERAL government plays a double role in the credit markets. It is not just an infamous borrower, but a major lender. In recent years, anywhere from a ninth to a fifth of all lending in the economy has been under federal auspices.

The trouble is that this lending activity, though a major cost and major part of the government's influence on the economy, is poorly accounted for in the budget. Direct federal loans are counted the same as expenditures when made, even though most of the expenditure is going to be recovered. Loan guarantees, on the other hand, don't show up in the budget at all when entered into, even though some are going to cost considerable sums of money when the borrowers default down the road. Then there are loans by so-called government-sponsored enterprises -- public-private hybrids -- which the government fosters but the budget never records.

At budget time these sharp differences in the treatment of similar activities have led to precisely the kinds of games you might expect. If you shift a program from direct loan to loan guarantee, the same volume of lending will occur with the same risk to the government, but you will get credit for a budget cut. You can sometimes get credit for a cut even if you increase the lending. Who wouldn't be for that? But it's a dangerous path, and part of the fine print in the new budget agreement between the president and Congress is a commendable effort to cut it off by rationalizing the accounting rules.

The new system won't reach the government-sponsored enterprises, a problem, insofar as they are one, left for another day. But it seeks to put loans and loan guarantees on an equal footing, and both on an equal footing with spending, particularly in competition for the incremental federal dollar. The idea is to carry loans and loan guarantees in the budget in terms of their estimated subsidy value -- how much it is expected they are ultimately going to cost the government. The currently overstated cost of loans would thus be reduced and the understated cost of guarantees increased. A committee -- or the administration, for that matter -- could no longer step up lending activity and have it be invisible in the budget; the true cost of increased loans would count against spending ceilings just like a spending increase, and Congress would be forced to decide which way to spend the next dollar in a way it is not forced to do now.

The major effects will be felt in those lending programs in which risks and/or subsidies are currently both high and masked. Congressional aides say these will likely include Farmers Home Administration lender-of-last-resort and rural development loans, and certain loans by the Rural Electrification and Small Business administrations. One of the tricks of the new system will be how to estimate true lending costs, and you can bet that that will produce its own round of attempted games. But it will be a better system than the one it replaces.