The dirty little secret embedded in the new federal welfare law is that it lets states replace welfare with…welfare. With AFDC gone, many states are making workfare the centerpiece of their anti-poverty strategy. It’s a big mistake. Welfare recipients shouldn’t be handed cash grants that they must “work off.” They should be offered wage-paying jobs. A good way to make sure they get decent jobs is to pay Employment Maintenance Organizations–which would function like HMOs–to move them into the labor market and keep them there.

The new federal block grant program, artlessly named Temporary Assistance to Needy Families, or TANF, no longer compels states to give destitute single mothers cash grants.

But bureaucratic inertia being what it is, most states will stick with cash grants with no work requirement for as long as the law allows. Other states will provide cash grants with a workfare requirement, as New York and Wisconsin are already doing. The idea is familiar territory and popular with voters.

But states should instead use their newfound freedom to create a wage-based system in which low-income parents who are able to work, but can’t find private sector employment, would be offered jobs. They’d get the minimum wage but they’d be treated like workers, getting a real paycheck and paying taxes. As wage earners, they would receive up to an additional $3,500 a year by claiming the federal Earned Income Tax Credit on their tax returns. Assuming, conservatively, that 200,000 people participated, that would pump as much as $400 million back into New York’s economy.

To replace welfare the right way, states should also take the program out of the hands of the bureaucrats. Bureaucrats focus on process. A successful program would focus on outcomes. It’s the difference between counting how many people go through a job training program and how many are working for a living.

The new welfare law includes goals for achieving minimal levels of work participation and offers incentives for cutting the welfare rolls. But it doesn’t seek to achieve minimal lengths of stay in employment. It has no earnings target. It offers no incentives for helping recipients get–and hold–higher-paying jobs.

But TANF doesn’t prohibit states from improving on the welfare law and demanding such solid outcomes. There are a number of effective ways a state could operate a results-driven, wage-based system. My favorite approach is what I call the “Employment Maintenance Organization” model, similar to the method used to buy from health maintenance organizations.

With this approach, states would establish a minimum earnings standard. For example, New York could aim to get 90 percent of its program’s participants into jobs paying at least $10,000 per year (excluding food stamps and the earned income tax credit.) States would randomly refer participants to Employment Maintenance Organizations (EMOs) and pay the organizations a fixed monthly fee per participant for a year.

Ideally, the EMO would get people long-term employment in the private sector. If an EMO fails to do so, however, it still must offer participants work–and make sure they are paid at least $10,000. Among the ways it could accomplish this are hiring them itself, placing them in a sheltered workshop or setting up a WPA-style community service program.

How much should the EMOs be paid per participant? That depends on the market. It could cost an EMO as little as $100 to find someone a job, or cost it as much as $12,000 to hire the person itself. An EMO that can maximize the number of participants it efficiently places in jobs is going to bida low price and be awarded more participants.

The program would thus include strong incentives for better service. If an EMO lost money because it did a lousy job of finding private employment for its participants, so be it. Fear of such a loss is exactly what would make the system work. It would motivate the EMO to meet the minimum earnings standards in a less costly manner by getting people into higher-paying private-sector jobs as quickly as possible.

If, on the other hand, an EMO makes good money by developing highly effective job placement methods, then it has served a valuable social purpose and deserves the profit.

To make the model even better, states could pay a bonus to any EMO whose participants have average earnings in excess of $12,000 for at least three years, and an even higher bonus if average earnings exceeded $14,000 for three years, and so forth. EMOs could reap additional profits proportionate to their outcomes.

None of this is to say that an outcome-driven, wage-based system would be perfect. Like any system, it would have to be well-managed to succeed. But compared to the process-driven workfare system many states are moving to, the model I’ve described would be a lot better.

It’s common sense. We don’t pay the roofer to try to fix the roof, we pay for a fixed roof. We don’t pay the baker for trying to bake us a cake, we pay for the cake. In most areas of economic life, we don’t pay for the attempt, but for the result. So why not cease paying bureaucrats for trying to help people move from welfare to work, and instead pay anyone who can do it to get the job done. Who, but those in the bureaucracy, could really object?

David Riemer is director of administration for Mayor John Norquist of Millwaukee. He helped develop the New Hope Project, Milwaukee’s test model of a wage-based welfare-to-work system.