DETROIT -- Sheila Thomas joined the welfare rolls in 2003 after her marriage fell apart, leaving her with three young children and no income. In order to collect payments of about $600 a month, the state required her to sign up for its welfare-to-work program.

Called Work First, the program routes Michigan welfare recipients to outside organizations that help with job searches and training. Its goal, like similar efforts around the nation, is to eventually wean individuals off public assistance.

Through Work First, Ms. Thomas has found jobs. The assignments, though, have all been temporary, each lasting just a few months. Today, she is no closer to achieving her goal -- or the state's -- of self-sufficiency. "I want to work," she says. "But the jobs keep ending."

Proponents of welfare-to-work argue that a client who gets a job, even a temporary one, is less likely to become chronically dependent on public aid. Nationally, between 15% and 40% of all welfare recipients who do work are in temp jobs.

The prevailing view among Detroit officials and others has been that temp jobs, despite their obvious disadvantages, help welfare recipients taste the dignity of work and develop valuable habits, such as punctuality and learning how to interact with others.

But a recent study of job programs in Detroit, which has one of the largest welfare populations in the country, is challenging the long-held belief that a temp job is better than no job.

"Encouraging low-skilled workers to take temporary help agency jobs is no more effective -- and possibly less effective -- than providing no job placements at all," says economist Susan Houseman, of the W.E. Upjohn Institute for Employment Research, a Kalamazoo, Mich., think tank. She is co-author, with David Autor of the Massachusetts Institute of Technology, of a study that tracked 23,000 Detroit welfare recipients.

The results surprised even the researchers. "If anything, we thought that temporary agencies would help welfare workers build skills, connect with potential employers, and so increase their future earnings," says Mr. Autor. "But this is not what we found."

Temp-agency work, they discovered, can create an unyielding cycle of finding and losing jobs. Detroit's Work First clients often had low morale, slim chances for job stability and plenty of setbacks. "While you're working at the temp job you're not connecting with direct-hire employers...you're not making any advances towards finding a permanent job," says Ms. Houseman.

The academic research confirms what some people involved in Detroit's system have suspected all along: that temp work has serious drawbacks. "Having a job this week and no job next week is not conducive to independence," says Melvin Chapman, director of a nonprofit welfare-to-work program called Diversified Educational Services Inc. Founded by his father in 1987, DES provides job training and placement assistance for 2,000 welfare recipients annually and generally shuns temp gigs for its clients.

One reason: When a welfare recipient loses a job -- especially through no fault of his or her own -- it can create resistance to wanting to work, says Mr. Chapman, a psychologist. In those cases, battle-scarred clients "are more difficult to send on another placement, saying, 'I know how this works, it's a merry-go-round,'" he says.

In addition to their ephemeral nature, temp jobs can have what the researchers call a "displacing" effect. They take up valuable time that welfare recipients could be using to find better, more stable positions. The temporary path may also divert some workers to dead-end jobs that ultimately discourage them from staying in the job market.

In Michigan, nearly half of the state's welfare recipients who exit the welfare system boomerang back within three months, says Marianne Udow, director of Michigan's Department of Human Services.

Detroit welfare officials say that while the preference is to find long-term jobs, temporary placements still are right for some. "Permanent is always the goal," says Deborah Watson, manager of the city's work-force development department. Finding such positions is "more challenging, yes, it is, but we do the best we can."

Although her office participated in the economists' study, Ms. Watson says she hasn't seen the results and could not comment on the findings.

Figuring out how to get people off welfare remains a significant issue long after President Clinton's welfare-reform laws were enacted a decade ago. His administration changed decades of government policy, limiting the time a person can remain on welfare to about five years and requiring that most people on cash assistance work, seek work or get job training.

The Bush administration recently signed a law updating welfare rules that will push states to move even more welfare recipients toward work and job training. The law requires that states place 50% of all their welfare cases into "work activities" and restricts what qualifies as work. Many college classes, for instance, will no longer count toward the work requirement as they once did. Credit for drug- and other substance-abuse programs is also limited.

Detroit's Work First program randomly assigns its clients to private job-placement organizations. Some send workers to temporary jobs; some don't.

In their study, conducted from late 1999 to mid-2003, Mr. Autor and Ms. Houseman compared work histories of individuals who went the two routes -- as well as those who tried to find work on their own. Of the individuals they studied, about 38% were placed in a permanent job during their Work First participation, 10% were placed with a temporary agency. The remaining 52% left the program without being placed in a job -- often because they didn't fulfill certain requirements such as attending mandatory meetings or submitting paperwork. In those cases, many clients also lost their welfare benefits.

Crunching data from the state's unemployment-insurance wage records, the researchers uncovered some surprising trends. Among workers with similar backgrounds, those placed in temp jobs earned more money at first than those who had been hired directly by an employer for a permanent post. But after a year, the temp workers were earning less money and had less stable employment. They were also more likely to wind up back on welfare than those who had been hired directly by a company into a permanent job.

Even more striking, workers who got jobs through temporary agencies over a two-year period earned about $2,200 less than those who didn't get placement but presumably found work on their own. The researchers' conclusion: Most welfare clients who were encouraged to take temporary help jobs by the Work First program would have done better by finding a job directly with an employer after some time spent searching on their own.

In 2003, Sheila Thomas, 46 years old, got a part-time grocery-store job paying $6 an hour through Work First. After the store closed in 2004, she was routed back to Work First, which sent her to a temporary agency.

For about four months she stuffed envelopes and put together in-store displays for a company specializing in paper products. She earned as much as $7.50 an hour. When that job ended, Ms. Thomas set off on her own to find a new gig through another temporary agency unaffiliated with Work First.

"I would sit there and wait and sometimes they had a job and I'd work a few hours," she says. "Sometimes they had nothing and I'd go home."

Ms. Thomas still longs for lasting employment. "Temp work isn't good because you can't look for something more permanent," she says, echoing the findings of the economists. "A temp job is 6 a.m. to 6 p.m. sometimes and so you don't have time to look. By the time you try to put your foot on the pavement, they're closed and looking at you like, 'Why weren't you here earlier?'"

To continue to qualify for aid, she is participating in job training and is trying to find work in the food-service business.

For the past two decades, Mr. Chapman's Diversified Educational Services has been trying to locate permanent jobs for people like Ms. Thomas. Although DES doesn't typically send clients to temporary-agency jobs, it makes an exception where an employer uses temporary positions as a probationary period, and puts workers on the permanent payroll after 90 days. The organization has a 60% placement rate and a track record that puts it in the top third of Work First placement providers.

Detroit Hispanic Development Corp., another Work First placement contractor, reluctantly sends some clients to temporary agencies. "We want to use them as our last resort," says Linda Gonzalez, the organization's director of employment and education. "But if it's something where time is running out, when all resources are exhausted, then we work with some of the temp agencies."

Some workers welcome temp assignments for their flexibility. The temporary-help sector has more than doubled over the past 15 years, to 2.6 million in 2006 from 1.1 million people in 1990. On average, most temporary jobs last three to four months, says Steve Berchem, vice president of the American Staffing Association, a trade group. He notes that temp jobs can lead to full-time positions or provide workers with marketable skills and training.

Mr. Berchem largely discounts the economists' findings, saying that other studies show temporary work does help boost low-skilled workers' wages over time. But none of those other studies used the same technique that Mr. Autor and Ms. Houseman used to compare experiences of similar individuals who were randomly placed in different work situations.

The emphasis on permanent, over temporary, work does seem to pay off for welfare clients like Tamra Fleming. A 29-year-old single mother of two, she began receiving cash assistance after having her first child at age 17 and was on and off the program for 12 years. Her first husband ended up in prison; she separated from her second six years ago. Ms. Fleming has been through Work First three times. Her initial stint resulted in a $6-an-hour, six-week summer job at White Castle. Her second led her to Spherion, a temporary agency, which sent her to do mostly light manufacturing work for auto suppliers.

After her third referral to Work First in August 2005, she ended up at DES, Mr. Chapman's outfit. First it sent her to a three-week customer-service training class. When it ended, a counselor lined up a job interview, advising her on what to wear and how to get a free outfit from another local nonprofit.

In October 2005, Ms. Fleming, who didn't finish high school, landed a $6.50 an hour job with President Tuxedo, a Detroit clothing store. In July 2006, she was promoted to store manager with a $35,000-a-year salary and health benefits. She bought a new house and a new car and has since started her own business as a wedding coordinator on the side. She is no longer receiving cash assistance.

Ms. Fleming says she wouldn't have found a permanent job without the extra boost from DES. "The welfare program should be pushing people a lot more," she says.

In Michigan, where 79,000 families are currently on welfare -- the state is in the process of replacing the Work First program. Ms. Udow said state officials wanted to change the program because they realized the notion of "finding a job, any job" wasn't working. "Our goal was to create permanent labor-force attachment and to reduce poverty in the state," says Ms. Udow.

The new initiative, called Jobs Education & Training, is being tested in four Michigan counties. It aims to help people find permanent jobs and stay employed. Unlike the Work First model, which immediately sends individuals out to find jobs, clients are first evaluated to size up their overall employability -- from labor skills to other factors such as child-care issues or problems with substance abuse.

As part of the program, Michigan is also working with several companies and nonprofit organizations that help welfare clients find -- and hold -- jobs. One of them is Cascade Engineering, a Grand Rapids-based manufacturer of plastic components used by auto companies and other industries.

Job candidates need to have at least 10th-grade reading and math skills, and must also pass a probationary period. After that, they are hired permanently at entry-level salaries of about $9 an hour. So far, Cascade officials say, most of the welfare clients sent to them have graduated to full-time positions. The turnover rate among those workers is low, at just about 2.5% annually.

Reprinted from The Wall Street Journal of Friday, December 16, 2006.