A decade after they became so obvious on park benches, at subway entrances, in front of stores, sleeping on grates, the homeless remain one of the most controversial groups of Americans.

There is still little agreement on where they came from. Even their numbers are in dispute, with estimates ranging from 250,000 nationwide (the Reagan administration) to about 3 million (the contention of some advocacy groups).

One popular explanation of their origin is that there has been a vast increase in the number of mentally ill people and of alcoholics and drug abusers.

Another is that the "deinstitutionalization" of mental patients dumped many into the streets on the assumption they would get community mental health services.

A third common theory is that subsidized housing cuts in the Reagan administration sharply reduced shelter available to poor people.

Now Martha Burt, an Urban Institute social researcher who has been studying the homeless in more than 100 cities for a decade -- even searching them out at night under bridges and in lightless hideaways -- has come up with some theories of her own in a book called "Over the Edge," to be published shortly by the institute and the Russell Sage Foundation.

She concludes that the three popular theories are myths. There has been too much emphasis on the "personal pathologies" of homeless people and not enough attention to structural problems in the U.S. economy that are the real causes of homelessness, she says.

According to Burt, the major reasons for the sudden visibility of homeless people are that low-cost housing became less available in the early 1980s and the social and economic structures that normally sustained people with very low incomes or with psychiatric, drug and alcohol problems also have declined.

Part of the picture, she says, is general economic change -- lagging wages, high unemployment -- as well as housing shortages that were not caused by the Reagan cuts. Another part is shrinkage of social programs at all levels of government.

People who are already poor and may have a drinking or drug problem, she notes, are always close to the edge. When such a person suddenly loses a job or the rent goes up, it is often enough to push him or her over the edge -- deeper into drinking, out of the home and onto the streets. And when some of the social services that might have seen such a person through the crisis are harder to come by, the situation is worsened.

Burt gives this composite picture: John, in his 40s, had worked for years in a factory, had a family, supported three children, drank a lot. But the factory closed during the recession of the early 1980s. He looked for work, drank more, gradually began leaving town looking for work. Eventually he stopped coming back. Now he picks up day labor jobs when he can, Burt says, "sleeps in cheap hotels when he can afford it and in shelters when he can't, eats at soup kitchens and continues drinking."

Homelessness, Burt found, increased enormously starting in the 1981-1982 recession when many people like "John" lost their jobs. At the start of the decade there were fewer than 250,000 homeless people, perhaps as few as 100,000. But by 1987 the number had grown to about 600,000, according to Burt.

About 73 percent of the homeless are single men, 9 percent are single women and 9 percent are women with children, Burt says. "Virtually all were very poor before they became homeless."

About half had spent time in mental hospitals or received chemical dependency treatments, and about half had spent time in a jail or prison.

Myth 1. Burt looked first at the idea that the big change in the 1980s reflected an upsurge in the number of mentally ill and drug and alcohol abusers.

"Population rates of chronic or severe mental illness are very stable across time and place, including non-Western and less-developed countries and cultures," Burt says. She found no statistical evidence of a sudden upsurge. "The proportion of the general population who are chronically mentally ill has not changed over the years, or in the 1980s," she says.

Alcohol abuse went down, according to most surveys. And while drug abuse increased during the late 1980s, that would not account for a rise in homelessness in the early 1980s, Burt says.

Myth 2. While many of the homeless were discharged from mental hospitals to the streets, "deinstitutionalization" cannot explain the dramatic rise in homeless people in the 1980s, because "the reduction of state hospital capacity was essentially complete by the mid-1970s," Burt says.

Myth 3. While Reagan's budget cuts for low-income housing did reduce the construction of future low-income rental units, she said, the dollars spent actually reached more households in the short run because they were used to subsidize more households in existing housing rather than to build new housing.

Until the 1980s, Burt says, most who would become homeless lived in housing of their own -- cheap flats, cubicle hotels, single-room occupancy (SRO) residential hotel rooms -- even many of the mentally troubled and substance abusers.

So, what drove them out onto the streets?

A primary explanation, Burt says, is a loss of affordable housing, though not primarily because of Reagan program cuts. During the 1980s, high interest rates for construction and tax law changes of various types sharply cut commercial production of low-cost rental housing. With higher costs for construction and also the bulldozing and closing down of many SRO units for urban renewal and highways, or for gentrification into fancier properties, many low-income people simply could no longer find affordable housing. Their only choice was the streets.

According to one study Burt cited, there was a "deficit" of about 4.9 million rental units at prices that the poorest 20 percent of households could afford in 1990 -- twice the deficit in 1970.

In addition, the 1980s have witnessed lags in wages for non-supervisory, non-professional workers, particularly in the poorest 20 percent of the population. Wages not only failed to improve in after-inflation dollars, but even lost ground. In 1979, weekly earnings in private industry (measured in 1977 dollars) averaged $183.41. At the end of the 1980s they had fallen to $166.52, according to Labor Department statistics.

And in the early '80s, especially the 1981-1982 recession period, unemployment was extremely high and poverty rose sharply -- from 11.7 percent of the population in 1979 to 15.2 percent in 1983. It was still 13.5 percent in 1990. Loss of manufacturing jobs in the cities hit many groups hard. And in some areas, wide gaps in income allowed better-paid workers to outbid the least-well-paid for what moderate-cost housing did exist.

While these developments were taking place, Burt says, many federal and state benefits in programs for the poor were failing to keep pace with inflation. These included Aid to Families With Dependent Children, state supplemental income payments for the aged, blind and disabled and state general assistance for the needy. Other programs, including unemployment insurance, raised eligibility and housing co-payment requirements, leaving the family or individual with less. Some individuals who had been able to struggle along in their own homes, even with really low incomes, mental health or substance-abuse problems, lost their ability to do so.

So, what is to be done?

Burt urges less emphasis on the "personal pathologies" of the people who become homeless and on stopgap solutions for individuals -- and more on making structural changes in the economy to end the conditions that push fragile people into homelessness. She says this means more housing subsidies to renters, but targeted at the very poorest, not at those who are not quite so poor. It means producing more housing that can be rented to the lowest-income groups even if it cannot pay its own way without subsidies. And it means much more education and job training and efforts to increase productivity in the economy, plus improvements in cash welfare programs to keep pace with inflation.

"If we succeed at this much larger agenda," Burt concludes, "we will not only solve the problem of homelessness, but also the problem of declining living standards for a much broader spectrum of American workers."