Randall Filer is a professor of economics at Hunter College and the Graduate Center of the City University of New York. This article is based in part on a larger research project conducted jointly with Professor Marjorie Honig, but the analysis and opinions expressed here are Professor Filer’s alone. Thanks are due to Nancy Degnan and Geoffrey Warner for research assistance.
To many New Yorkers their daily, depressing, and sometimes infuriating contact with the homeless in the city’s public places suggests a problem of crisis proportions. In fact, New York’s particular homeless crisis is not found among the mostly single men who live on the street. Such people are actually less common here than in other U.S. cities.
Where New York stands alone is in the extent of homelessness among families, which is more than twice as common here as in most other large cities. That is a disturbing figure. Even more disturbing is evidence that the city’s homeless policies have caused much of the problem by creating powerful incentives for poor families to become homeless.
This is not the usual explanation for New York’s exceptional rate of family homelessness, or for its steady increase throughout most of the 1980s. Two plausible explanations are commonly advanced. First, New York’s reportedly tight and expensive housing market may push more families into homelessness. Second, a greater than average fraction of New York residents may have characteristics (such as poverty or being single heads of households) that increase the risk of becoming homeless. A close look at the evidence, however, suggests that these oft-cited factors explain but a small portion of New York’s problem.
The search for an explanation starts with comparing New York City to 20 other cities for which reliable data on rates of homelessness are available.1 The key is to compare these cities to New York not only with respect to homelessness, but also with respect to the factors generally advanced to explain New York’s high homelessness rate: housing availability and affordability, poverty rates, and prevalence of single heads of families.
The homeless in New York
The data show that there were approximately 1.4 homeless single adults in New York City for every 1,000 residents in 1984, compared to an average of 3.25 per 1,000 in the other 20 cities. In other words, the rate of homelessness among singles in New York is only about one-third that in the other cities.
For homeless families, the situation is almost exactly the reverse. There are 2.67 homeless members of families in New York City for every 1,000 residents, compared to an average of only 1.15 homeless family members per 1,000 residents in the other cities. Since homeless families in both New York and other cities have between 3 and 3.5 members, the same relationship holds for homeless families as for homeless members of families.
In many ways homeless families in New York are similar to homeless families in other cities. They overwhelmingly consist of single women and their children. Between half and two-thirds of these women have not graduated from high school and few have, any attachment to the labor market. Homeless families in New York, however, do have substantially larger monthly cash incomes than their counterparts in other cities: $422 versus $238, according to a 1988 survey by the Urban Institute. In large part, this is because New York’s homeless families are more likely to receive welfare payments. Eighty-six percent of New York’s homeless families were receiving food stamps, compared to only 31 percent of homeless families in other cities. Sixty-five percent were receiving General Assistance (GA) welfare payments, as opposed to 16 percent elsewhere, while 38 percent of homeless families in New York were getting income from Aid to Families with Dependent Children (AFDC), compared with 31 percent in other cities. That more of New York’s homeless families receive welfare benefits suggests that they are more representative of the general low-income population than homeless families in other cities who may have fallen through the cracks of the social support system.
New York’s housing market
To see if New York’s tight housing market is the cause of its unusually high rate of family homelessness, the affordability and availability of apartments in New York can be compared to the situation in other cities. Further evidence can be obtained by comparing changes in family homelessness to changes in the housing market.
Whether housing is affordable depends on the relationship between housing costs and incomes. It matters little if rents in a given city are high if incomes are also high there. To determine the affordability of housing for the potentially homeless, rents at the low end of the market should be compared with the welfare benefits that provide most of the income of potentially homeless families. (In the Urban Institute survey, fewer than three percent of the heads of homeless families in New York had any recent labor income.)
Figure 1 shows the relationship between apartment affordability for the poor (the ratio of rents for apartments at the 10th percentile of the rent distribution2 to the maximum AFDC benefits plus food stamps for a family of three) and the extent of family homelessness3 in the cities for which we have data. The results are the same if any point in the lower tail of the rent distribution is used. It is clear in Figure 1 that there is no relationship between housing affordability and family homelessness.
Strikingly, housing is no less affordable for poor New Yorkers than for poor people in the country as a whole. In both New York and elsewhere it would cost a poor family of three about a quarter of their income to rent an apartment at the 10th percentile of the rent distribution.
It would not matter, however, whether average rents were affordable in a city if no low-rent apartments were available. Average rents could be lowered by inexpensive occupied apartments while most vacancies occurred in expensive units. Thus, a better indicator of affordability is the price of vacant apartments. When the affordability analysis is repeated using rents at the 10th percentile of vacant apartments, housing in New York turns out to be even more affordable for the very poor than in other cities. In 1984, a family of three with no earned income would have paid 24 percent of its AFDC and food stamp benefits to rent an apartment at the 10th percentile of vacant units in New York, compared to an average of 28 percent of its income for such an apartment in the other 20 cities.
Finally, we must ask whether the problem with housing markets in New York is the sheer availability of apartments rather than their affordability. After all, the overall vacancy rate for New York City is about half that for the other cities in our sample (3.2 percent versus 6.1 percent). The same relationship holds in low-cost housing units, with 3.4 percent of apartments below the 10th percentile of the rent distribution vacant in New York, according to the 1980 census, compared with an average of 6.2 percent elsewhere.
Vacancy rates alone, however, are not particularly informative.4 What matters is the number of low-cost vacant apartments relative to the number of people searching for such an apartment. The ratio of homeless families to low-cost vacant units is in fact higher in New York than in the other cities for which we have data. There are slightly over five vacant apartments at or below the 10th percentile of the rent distribution for every homeless family in the other cities versus fewer than one and a half such units in New York.
It is important, however, to untangle cause from effect. If there are an unusually high number of homeless families in New York for reasons other than the housing market, this would create a low ratio of available units to homeless families. To isolate the effects of housing availability on homelessness, it is more appropriate to compare the number of low-rent vacant units to the number of homeless families New York would have if it had the same rate of homelessness as other cities.
If New York’s homelessness rate were the same as in the other cities, there would be 4.91 vacant apartments at rents lower than the 10th percentile for every homeless family in New York compared with 5.04 such vacant apartments elsewhere. This tiny difference cannot have created the massive difference between the number of homeless families in New York and other cities.
Other evidence points to the same conclusion. When poor families cannot find apartments, they often share housing with friends and relatives. Indeed, the vast majority of families entering the homeless shelter system come from shared housing. If the rental market were tougher for poor families in New York than in other cities, there should be more doubling up among the poor here. Once again, however, the difference between New York and the other cities is minimal. In the 1980 census, 8.1 percent of New York households in which the head of the household’s income was below the poverty level included more than one nuclear family. In the other cities to which we have been comparing New York, 7.5 percent of such households were doubled up. Results are similar for female-headed households. Twenty-five percent of these households contained more than one nuclear family in New York, only 1 percent more than the 24 percent doubled up elsewhere. Housing is difficult for the poor everywhere, but no more so in New York than in other cities.
Finally, if a shortage of low-cost housing is responsible for the high rate of family homelessness in New York, there should be a relationship between fluctuations in the housing market and fluctuations in the number of homeless families. Such a relationship does not exist.
Figure 2 shows the average number of homeless families sheltered by the City of New York each month from January 1983 through March 1990. One does not need a course in statistics to see the pattern. The number rose at a steady rate through the middle of 1987, when slightly over 5,000 families were being sheltered by the city, then leveled off for a year before beginning to fall. By the beginning of 1990 the number of families sheltered by the city may have stabilized at about 3,800.
Two time periods are especially telling. The number of homeless families in New York increased steadily from 1983 through mid 1987. Yet in January of 1984 the shelter allowance provided AFDC recipients was increased by 25 percent. If housing affordability were a major contributor to homelessness, one would expect a change in the trend line in Figure 2 after poor families received this increase in purchasing power. Instead homelessness among families continued to rise at the same rate.
In early 1987 this trend was reversed. The number of homeless families in New York declined through 1988 and early 1989. Nothing happened to rents or welfare payments in early 1987, however, to significantly change the housing market for poor families.
New York’s poor
If housing markets do not explain either the level of or trends in homelessness among families in New York, what about the second commonly advanced explanation? Does New York simply have more families with characteristics that put them at risk of becoming homeless?
Two major, related contributors to homelessness are poverty and female-headed families. Both of these are somewhat more common in New York than in our comparison cities. In the 1980 census, 20.1 percent of New York households were headed by women as opposed to 15.5 percent in the other cities. There was a similar difference in the incidence of poverty, with 16.3 percent of household heads in New York reporting cash incomes below the poverty level, compared to 11.1 percent elsewhere. This 30 to 50 percent higher proportion of families at risk can explain only a small portion of New York’s 250 percent greater rate of family homelessness.
Family characteristics also fall to explain the changes in the number of homeless families over time. There is no evidence that poverty or female-headed households became vastly more common in New York between 1983 and 1987, or less common following 1987.
New York’s policies
It is time to examine the third possibility. Perhaps the major cause of New York’s unique homelessness problem is New York’s housing and homeless policies. These policies favor the homeless over the housed poor in a number of ways and are significantly more generous towards the homeless than those in other cities. Since more poor families can improve their lot by becoming homeless in New York than in other cities, it should not be surprising that there is a higher rate of homelessness among families in New York.
For families in New York, “homelessness” does not mean living on the street or in the subway. There are, in recent years, virtually no recorded instances of families living in public places. It would be surprising to find many unsheltered families in New York because the city has a legal obligation to shelter any individual or family who claims to be homeless and requests such shelter. This requirement has resulted from a number of court rulings and consent decrees, culminating in the case of McLain v. Koch in 1986. The city fulfills its obligation provide shelter by placing families in hotels of varying quality, by placing them in public shelters (which may range from congregate facilities with dormitories and shared public spaces to facilities offering quasi-apartments), or by finding them a permanent space in public housing. As we will see, poor families have strong preferences regarding these options.
In addition to those families who will become homeless no matter what the city’s policies—those who lose an apartment to fire, those whose husbands or boyfriends become abusive, those whose parents simply cannot deal with them any longer and so forth—there are many other families for whom homeless shelter provided by the city is one of a set of poor housing options. A majority of the families who seek shelter from the city have previously been sharing housing with friends or relatives. There may be as many as 100,000 other doubled-up households who could opt to enter the city’s shelter system if doing so would improve their housing options.5
Advocates call such families the “hidden homeless.” For them, as well as other very poor families who are currently renting apartments, the decision to enter the shelter system is a matter of comparing costs and benefits. They will seek shelter from the city when the benefits of doing so are greater than the costs. Thus, the fact that the city cannot turn away any family who desires city shelter does not mean that it cannot limit the number of sheltered families. The city can determine the number of families it shelters by the quality of shelter and the other public benefits it provides to homeless families.
Although it may seem odd to talk about the benefits of becoming homeless, these benefits are very real and may be quite substantial. In New York they consist largely of priority access to subsidized housing and increased disposable cash income.
Since 1984, New York City has given priority to homeless families for new or renovated housing owned by the city. Currently a poor family applying for subsidized housing in the ordinary way faces a waiting list containing in excess of 200,000 families. The size of this list implies an 18-year wait for such housing (although duplications and inclusion of families no longer eligible for or desiring subsidized housing means the actual wait will be less). Families entering the homeless shelter system, on the other hand, wait about a year for a subsidized apartment.
Moreover, becoming homeless can actually raise a family’s income. This is especially true if the family is able to secure placement in a hotel room rather than a city shelter. In 1987, for example, a family of four with no earned income would receive a basic AFDC grant of $326 a month, plus a housing allowance of $270 and $62 in food stamps. Assuming their rent did not exceed the housing allowance, they were left with $388 a month for other needs. If the same family became homeless and were assigned a hotel room, it retained its basic grant and food stamps of $388 a month. In addition, that family would get a “transportation allowance” to enable the mother to accompany her children to school in their old neighborhood, plus a “restaurant allowance” of $362 a month because hotel rooms do not typically have kitchens. Since homeless families in New York’s hotels prepare over 90 percent of their meals in their rooms anyway, the “restaurant allowance” actually increases their income by over $350 a month. This nearly doubles their disposable cash income.
Homeless families placed in city shelters, on the other hand, get a transportation allowance but no restaurant allowance or food stamps because the shelters provide meals. Whether entering a shelter increases their incomes depends on whether the mothers use their travel allowance to accompany their children to school, on whether they had been spending more than their food stamps for food, and on the value of the meals provided by the shelters.
The effect on income is especially pronounced for one group of potential homeless families—young women with children who have the option of living with their parents. New York assumes that when a woman lives with a parent, the parent provides room and board and reduces her welfare payment accordingly. Thus, a single woman with a child will increase her income if she leaves her family and becomes homeless. The city, instead of her parents, provides her with food and shelter, but she will not have to make any contribution to household expenses. Her base welfare grant increases because the city no longer assumes that she is being subsidized by her parents. New York is one of only 10 states that lower a woman’s welfare if she decides to remain at home. In 28 grant states plus the District of Columbia, she receives her full grant (including shelter allowance) no matter where she decides to live, while in the remaining 12 states she can retain her full grant by making a token payment of as little as a dollar a month to her parent(s) for room and board.
The costs of becoming homeless are the conditions under which families live while in the shelter system. These costs vary greatly depending on whether families are placed in hotels or city shelters, with hotels being by far the more attractive option. We know hotels are the more desirable alternative because homeless families say they prefer hotels and because they behave in ways designed to increase their chances of being assigned to them.
Among families applying for city shelter in December of 1985, 61 percent said they would accept only certain types of shelter. Over half of the families would decline a congregate shelter, but would accept a private apartment or hotel room, according to a report by the New York City Human Resources Administration (HRA). According to another HRA report, among residents of one welfare hotel in February 1986, 46 percent would refuse placement in a shelter with private bedrooms but shared kitchens and recreation areas, and 61 percent would refuse shelter that provided private bedrooms but did not allow non-family members to visit in those bedrooms.
The timing of requests for shelter by homeless families shows their preference for hotel rooms. During normal business hours homeless families should report to an Income Maintenance Center (IMC) located in their neighborhood for assignment to shelter. Those who become homeless at other times can go to Emergency Assistance Units (EAUs) between 4 P.M. and 8 A.M. on weekdays and 24 hours a day other days. Although the city originally expected almost all assignments to be made through the Income Maintenance Centers, in the mid 1980s almost three-quarters of family requests for shelter came from Emergency Assistance Units.
In part this was because families not placed by their IMC reappeared at an EAU. The primary reason, however, seems to be that only 10 percent of the families at IMCs were assigned to hotels, while fully half of the placements from EAUs were to hotels. Some families refused the placement offered by their IMC and tried again at an EAU, while about half of those appearing at the EAUs had not bothered to go to their IMC at all. Many families appeared at an EAU several times, refusing the shelter offered until they were placed in a hotel. Administrative records show that about one-quarter of the families offered shelter by the city (almost all of them families who were not offered hotel rooms) made their own sleeping arrangements rather than accept the shelter provided.
The percentage of families who refuse shelter may be even higher. During 1985, the city received about 5,000 requests for shelter each month from families, according to Steven Taylor, then Deputy Administrator for Policy and Program Development at the Human Resources Administration. Eliminating duplications, at least 2,500 families were referred to hotels and shelters. However, only 600 families entered the system. The other 1,900 must have made alternative arrangements.6
According to Robert Jorgen, in 1984 the head of the Human Resources Administration’s Crises Intervention Services Unit, “When the city was still sending people to New Jersey hotels we got tremendous publicity. We had some fairly nice hotels out there. People saw on TV all the kids splashing around in the swimming pools. Our EAU was packed with people wanting to go to New Jersey who turned around and left when they were told they couldn’t.”7
More evidence of poor families’ preference for hotel rooms comes from the behavior of pregnant women. In 1988, over one-third of the heads of homeless families applying for shelter were pregnant. By comparison, only 6 percent of heads of housed welfare families are pregnant at any time.8
Why should pregnancy be so much more common among women requesting shelter from the city than among other poor women? It is unclear how a future birth could strain current housing resources. One might theorize that unwanted pregnancies lead fathers to throw the mother out, but very few heads of homeless families have recently lived with either a husband or boyfriend.
The answer lies in city policies. Pregnant women face a different cost-benefit calculus than other women. Pregnant women (and new mothers) receive priority over other homeless women for permanent placement in subsidized housing. They are also guaranteed temporary placement in a private room or hotel. Given these lower costs and higher benefits, more pregnant than non-pregnant women enter the shelter system. Since a woman’s early placement in housing and better conditions in the shelter system are eliminated if she ceases to be pregnant, the city’s policies may also encourage women to decide to bear children they cannot support.
While we have no evidence of women getting pregnant in order to obtain priority for housing, it is clear that women understand and respond to the incentives in city policies. Ralph Da Costa Nuñez of Homes for the Homeless and Ed Abrams of the Food and Hunger Hotline both report that there is an active street market in pregnant women’s urine, bought by non-pregnant women who wish to secure private accommodations and jump to the head of the line for public housing. This happens often enough that the city has instituted special procedures to attempt to catch fake pregnancies. Surely such subtle behaviors designed to obtain their preferred treatment belie the claim that poor families are too unsophisticated to understand or respond to the city’s policies.
We can now begin to understand the forces that have created New York’s high level of family homelessness. Imagine a city with no homeless families but a waiting list for public housing. Now suppose the local government announced it would provide a fixed number of apartments to homeless families every year but that while waiting for placement in these apartments each family would have to live in a shelter of a particular quality. Low-income families would compare the benefits from obtaining priority for subsidized housing with the costs of living in the shelter until placed (including any effects on income and quality of life).
Obviously, if there were no wait in a shelter, all families desiring subsidized housing should declare themselves homeless and obtain immediate access to permanent public housing. If the wait for a permanent placement were short, and the temporary shelter pleasant, nearly as many families would declare themselves homeless. As the wait for placement increased, or the quality of the temporary shelter decreased, declaring homelessness would become less attractive until, at some point, the marginal family would no longer find it worthwhile to endure the deprivations of shelter life in return for earlier placement in subsidized housing. In this way the decisions of poor families themselves determine the “optimal” length of stay in the shelter system along with the number of families being sheltered.
Contrary to conventional wisdom, providing more permanent housing for the homeless would increase rather than reduce the number of homeless families. If the city built more housing for homeless families in an attempt to reduce the length of time each family spent in its shelters, this would induce more families to join the queue until the wait for shelter is restored to its equilibrium length (where the costs of time spent in the shelter system just equals the advantage of early placement). Thus, increases in permanent housing provided for homeless families should result in a proportional increase in the number of homeless families, but no change in the average period of homelessness.
This analysis predicts that there should be more homeless families in New York than other cities because New York provides substantially larger benefits to its homeless and houses them under substantially better conditions than other cities. In 1988, for example, New York spent almost $25,000 on an annual basis for each sheltered homeless person, compared with an average of $5,500 in the next four largest cities in the country (Los Angeles, Chicago, Houston, and Philadelphia).9 In fact, New York City spends seven times as much on programs for the homeless as the combined spending of the next four largest cities, even though their total population exceeds New York’s by a third.
New York is also unique in its efforts to provide permanent housing for the homeless. By some estimates it spends more on housing programs than all other American cities combined. In fiscal 1988, almost half of New York’s expenditures for homeless persons was for housing renovation, despite the evidence presented earlier that housing markets play very little role in the extent of homelessness in New York.
Since most cities neither provide restaurant allowances to homeless families nor reduce welfare grants to families who share housing with relatives, the increase in income that can result if a family becomes homeless is larger in New York than elsewhere.
The human costs to the homeless of seeking shelter are lower in New York than in other cities because it makes more extensive use of hotel rooms to shelter its homeless. The consequences of such a decision can be clearly seen in the experiences of nearby cities. In the fall of 1989, Westchester County housed the majority of its homeless families in hotels and gave them restaurant allowances while Nassau County sent its homeless families to shelters. Not surprisingly, Westchester was sheltering approximately 800 families, while Nassau (with a population about one and a half times as large) had only 50 homeless families.10 New Haven, which uses hotels, reported family homelessness rates six times those of Hartford or Bridgeport, which use shelters.
Further evidence that the number of homeless families in New York responds to the city’s policies comes from the growth trend in such families. Recall the pattern in Figure 2. The number of homeless families rose steadily until early 1987 when it reversed itself even though there was no significant change in either the housing market or family characteristics. What did change in 1987 was the city’s strategy for sheltering the homeless. At the end of 1986, over three-fourths of the families sheltered by the City of New York were placed in hotel rooms. In 1987, partially in response to Federal Government threats to cut off funding for hotel rooms, the city began to reduce its use of hotels to shelter homeless families with the goal of completely eliminating hotels from the shelter system by 1990 (since revised to 1992).
When this change in policy increased the costs of becoming homeless, the number of homeless families dropped. This occurred even though the number of permanent units provided for the homeless continued to rise, just as it had been doing when the number of homeless families was increasing. Thus, the length of time a family could be predicted to wait in the shelter system for subsidized housing fell from 15 months in fiscal 1986-1987 to 10 months in fiscal 1988-1989, just the result one would expect to see if changes in policy (fewer hotel rooms) had increased the cost of waiting, thereby reducing the equilibrium waiting time.
The real debate
We have seen that the exceptional rate of family homelessness in New York is not a result of housing-market conditions. Apartments are as available and as affordable for the poor in New York as they are elsewhere. There are, however, more poor and female-headed families in New York. This explains a portion of the difference in family homelessness between New York and other cities.
The bulk of the difference between New York and other cities, however, is best understood as the result of more generous rewards and lower costs to entering the public shelter system in New York. Building or setting aside more permanent subsidized housing specifically for the homeless will not solve the problem. Instead, it will make things worse. If more units are provided, there will be a matching increase in the number of families waiting in the shelter queue for housing as families act to balance the advantages of early access to permanent housing with the disadvantages of shelter life. Making the shelters themselves more pleasant will also increase the number of families willing to join the queue until a new, longer optimal waiting time is reached.
Programs for homeless families cannot be more generous than those for poor families in general without increasing the number of families who eventually endure a spell of homelessness. This is obviously not a desirable result. Even though families may find it to their advantage to use the shelter system to secure better permanent housing, such use involves real costs for the family as well as the city. Providing emergency shelter is an expensive way of screening applicants for housing assistance, thereby depriving other needy people of the city’s resources. Because poor families do understand and respond rationally to incentives, a commitment to provide housing for all homeless families is, by definition, eventually a commitment to provide housing for all poor families. Such a commitment should be debated on its own merits, and the debate not clouded by the emotional specter of homelessness.
- Until the release in a few years of the data from the 1990 census, the best information on the number of homeless people is provided by the U.S. Department of Housing and Urban Development, which estimated the total number for various cities in 1984. Independent evidence suggests that the HUD estimates were highly accurate. The number in families is obtained by multiplying this total by the proportion of homeless persons in families estimated by each city for the U.S. Conference of Mayors in 1985 and 1986. The 20 cities are all those for which data are available in both sources: Boston, Charleston, S.C., Chicago, Cleveland, Detroit, Hartford, Kansas City, Los Angeles, Louisville, Minneapolis-St.Paul, Nashville, New Orleans, Philadelphia, Phoenix, Portland, Ore., Providence, Salt Lake City, San Francisco, Seattle, and Washington, D.C.
- The distribution of rents is only available at the city level in 1980 census data. They have been updated to 1984 by increasing the 1980 rents in each city by the rate of increase in rents between 1980 and 1984 for the nation as a whole. Thus, results for any individual city will be biased by the extent to which that city’s change in housing prices for these four years differed from the U.S. as a whole. In the early 1980s, real housing prices in New York City increased at about the same rate as the national average.
- Because the fraction of the metropolitan area contained within the central city varies across cities, Figure I uses metropolitan area population in the denominator rather than the central city population used to calculate the raw numbers presented above. Results are the same no matter which measure is used.
- Rigidities introduced by rent control mean that a large fraction of apartments in New York are held by longterm tenants and are not part of the active rental market. In addition, a low vacancy rate need not indicate a poorly functioning housing market. Optimal inventory theory predicts that the ratio of inventory to sales (the vacancy rate) will be lower the larger the scale of operations. Indeed, New York’s vacancy rate is actually higher than would be predicted from the relationship between vacancy rates and the size of the rental market across other cities.
- A Shelter is Not a Home: Report of the Manhattan Borough President’s Task Force on Housing for Homeless Families, by Jame R. Dumpson, Chairman (New York, NX: Office of the President of the Borough of Manhattan, 1987).
- Cited in Thomas Main, “The Homeless Families of New York,” The Public Interest, Fall 1985, 18.
- Cited in Tom Robbins, “New York’s Homeless Families,” City Limits, November 1984, 8.
- James R. Knickman and Beth C. Weitzman, “A Study of Homeless Families in New York City,” (New York, N.Y.: New York University, Graduate School of Public Administration, 1989) mimeo.
- U.S. Department of Housing and Urban Development, Office of Policy Development and Research, Homeless Assistance Policy and Practice in the Nation’s Five Largest Cities ([Washington, D.C.]: U.S. Department of Housing and Urban Development, 1989).
- Robert C. Ellickson, “The Homelessness Muddle,” The Public Interest, Spring 1990, 48.