WILLIAMSBURG -- If there is any consensus in American politics, it is that the present system of public welfare doesn't work very well. Both House and Senate, and liberals and conservatives, are laboring over the question of how best to go about fixing it.

According to Robert Reischauer of the Brookings Institution, the reformers may be asking the wrong question, or at any rate asking it prematurely.

The fundamental choice, he told a Rockefeller Foundation-sponsored conference on welfare reform here, is not how to fix welfare but whether to fix it. To the general approval of the assembled welfare experts, Reischauer suggested that it might make more sense to focus on nonwelfare approaches to improving the plight of poor Americans.

"The welfare system exists largely to pick up the shards left when other systems and institutions fail," he told the conference, making clear that the views expressed were his own and not necessarily those of Brookings.

"Failures in the educational system, in marriages and families and the labor market are the major systemic causes of welfare dependency. Resources can be devoted to strengthening these institutions and systems in the hope that this will reduce the long-run burden on the welfare system."

For example, he said, it might make more sense to try directly to reduce the incidence of out-of-wedlock births (linked to nearly a third of new welfare cases) than to try to use welfare reform to do the job. "Similarly, more of an effort could be made to strengthen the educational systems that serve youth from low-income families. This might mean increasing the budgets of the Head Start and Chapter 1 programs or pursuing more radical educational reforms such as year-round schooling for disadvantaged children."

Raising the minimum wage, expanding job-training programs and removing such barriers to employment as inadequate day care and lack of health insurance might do more to lift welfare recipients out of their poverty than any of the "workfare" provisions of the pending reform legislation.

While there is room for debate over particular nonwelfare approaches (conservatives, for instance, argue that increasing the minimum wage is likely to reduce the number of entry-level jobs), the general thrust of Reischauer's argument makes so much sense that many of us are likely to think that it is what we always believed.

Well, we haven't always believed it. We have argued welfare reform as though taking it as a given that a properly devised welfare system would foster the sorts of attitudes and behaviors needed to lift the poor out of their poverty. Reischauer's contention is that other means of attacking poverty might be more effective, leaving welfare for the hard cases that seem immune to the more direct approach, "the shards left over when other systems and institutions fail."

Welfare, perhaps inevitably, involves a conflict between fundamental American values. We believe that we should provide enough income support for poor people to lift them out of squalor, but we also believe that to do so would increase the number of dependent people by eroding their incentive to help themselves.

Welfare reform, particularly its workfare requirements, seeks to overcome the dilemma by forcing recipients to earn their grants. But as Reischauer suggests, if we do what is necessary to help people earn their welfare grants -- improving their education and training, solving their child-care problems and guaranteeing them a decent wage -- we may find that welfare is unnecessary for all but a hard-core few.

In other words, the best approach to welfare may be to improve our nonwelfare programs.

Not only would such an approach be more effective; it might also be more sparing of tax dollars, since increases in health insurance and the minimum wage would be borne largely by the private sector. In addition, it might be less divisive, since it would reinforce mainstream values.

Reischauer wouldn't eliminate welfare. He might even improve it modestly, in line with most of the present "reform" proposals.

But he wouldn't look to welfare to improve the lot of most poor people. He wouldn't even argue that nonwelfare approaches would necessarily do the job.

As William Julius Wilson, the University of Chicago sociologist, noted in another paper presented at this Rockefeller Foundation conference, it is fruitless to look to welfare reform to cure what is at bottom a problem with the American economy.