THE REAGAN ADMINISTRATION is so anxious to dismantle the Labor Department's job and training system that, even before congressional approval, it has begun phasing out jobs now provided to low-income adults. Plans under consideration also call for eliminating funds for year-round youth job and training programs and consolidating what is left of the Comprehensive Employment and Training (CETA) system into "block grants" to states and localities. This would reduce support to less than half of the $9 billion planned by the Carter administration, making job programs the largest loser in the Reagan budget.

Job and training programs for low-income persons might seem like a curious target for an administration committed to encouraging work effort, improving labor productivity and reducing welfare dependence. A further irony is that, despite their well-documented shortcomings, job programs have, in fact, been shown by careful studies to increase the subsequent earnings of participants, produce numerous useful services to communities and cut welfare and unemployment costs. Recent efforts to move welfare recipients into permanent jobs and to reduce the number of school dropouts have been particularly promising, although these projects are already being terminated.

Each year CETA helps more than three million people get into the job market. This severe retrenchment, hard on the heels of earlier Carter administration cutbacks, will clearly cause a lot of long-term hardship.But the administration's enthusiasm for the "no-strings-attached" block grant approach is worrisome in itself. CETA had its origins in the similar "manpower revenue-sharing" proposals of the early 1970s, and a lot of its continuing problems stem from that very fact.

CETA is not actually a federal program at all -- simply a mechanism for letting grants and contracts to more than 40,000 state and local agencies, non-profit organizations and educational institutions that provide the actual jobs and training. Because it is run primarily by local government officials, it is no better and no worse than local government in general. When CETA programs run well, they are generally referred to by the name of the sponsoring organization -- the Community Uplift Program, the Youth Enterprise Initiative and so on. When something goes wrong with the programs, they are always referred to as "CETA."

Even before CETA's period of rapid expansion in 1977, it was amply clear that more federal control was needed. Left to their own devices, local officials used CETA money to hire and train many people they would have hired anyway to do jobs that were already being done. Since then, major reforms have succeeded in focusing CETA programs on their intended purpose of creating new job and training opportunities for the kind of people whom, as has been amply demonstrated, neither private employers nor the regular school system are willing and able to help. But along with these necessary controls came a lot of unnecessary red tape -- lists of special "target groups," set-asides for pet projects and the like -- added by Congress at the behest of interest groups.

By all means let us get rid of the red tape. But let's not give a free hand to local governments to undo all the careful but unfinished work of making job programs really help their intended clients. More, not less, of this kind of control is needed.

There is nothing sacrosanct about the current size of CETA. But there are much better ways to save money than those the administration is considering. Requiring a state and local contribution to job programs would both reduce rederal costs and give the people who run the programs a clearer stake in their effectiveness. And federal funding formulas should be changed to reduce money spent in many areas that don't really need it.

But if the administration really sees no national interest in improving job skills and opportunities for the needy, it should say so and get out of the business. This would be a sad and shortsighted decision, but it would be better than buying off shrieking states and localities with a still sizable chunk of money that is unlikely to do much for either labor markets or the poor.