THERE IS a curious inconsistnecy in the administration's approach to incentives for the rich on the one hand, and the poor on the other. The rich, we are told, must be given financial encouragement to maintain their work effort and expand their investments. The poor, by contrast, are to be encouraged to work by the threat of being made even worse off.

The administration plans to require able-bodied parents in families receiving Aid to Families with Dependent Children (AFDC) to "work off" their welfare grants in specially created public jobs. Recipients would not be paid wages. Instead, they would be given credits for the hours they work at a rate somewhere near the minimum wage. If they worked fewer than the necessary hours to pay off their grants, their benefits would be reduced.

We do not have to look to the heavily socialized countries of the world to demonstrate that exhortations to work or threats of reprisal are not effective ways to promote productive work. There are examples from home to prove the point.

In 1971, after a four-year period in which AFDC cases more than doubled, California began to experiment with work-relief projects of the type now being proposed for the country. The projects were not a success. Several of the largest counties refused outright to take on the job of supervising unwilling workers in makeshift, part-time jobs. During 1974, the peak year of the program, only about 5,700 people were enrolled -- less than 1 percent of AFDC cases in California that year. A study done by the state found no difference in trends in welfare costs and cases between counties with work projects and those without. AFDC cases did start to level off in California even before the projects started. But the halt in AFDC growth -- which also occurred in most other states over the next few years -- was caused by other factors, including a general tightening up of previously lax program administration.

In similar work-relief projects in Massachusetts, it was found that recipients enrolled in the projects actually stayed on welfare longer than a comparable group that was left alone.

Helping welfare recipients get permanent, decent-paying jobs is a humane and farsighted idea. And a work requirement can be a helpful prod to overcome the self-doubt and apprehension of welfare parents with little or no recent work experience. Various other pilot projects have show that good job-finding, work and training programs -- in which participants can increase their incomes by working -- will pay off for both taxpayers and welfare recipients.

The reason for the success of the latter projects, and the failure of the work-relief approach, is simple. Most welfare recipients want to work if they see employment as a dignified way to a better life. But it will come as no surprise to the supply-siders that the poor -- like, presumably, the rich -- are unwilling to work for little or no financial gain. Indeed, the only hard data we have on people's responses towork incentives come from the income-maintenance experiments that studied poor and near-poor families.

Herding unwilling people into make-work jobs won't work, it won't save welfare money and -- in the long run -- by cheapening the idea of work for welfare parents and their children, it will increase welfare dependency.