Inner-City Poverty in the United States (1990)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

How Poverty Neighborhoods Are Changing JOHN C. WEIGHER Public concern about urban problems has tended to focus on the notion of an "underclass" during the 1980s. The concept has not been precisely defined, at least to the point where analysts have general agreement on its meaning and composition, but it is perhaps fair to say that the underclass includes people who are persistently poor, especially those who also grew up in poor families, and who live in neighborhoods where much of the population is persistently poor. These neighborhoods might be termed "persistent slums." This chapter provides a microanalytic perspective on the phenomenon of persistent slums as a contribution to better understanding of the problem. It describes the changes that have occurred in selected urban poverty neighborhoods during the 1970s. The focus of the analysis is on the private characteristics of the slums: the people and the housing~what they are like, and how they are changing. The chapter also looks at how federal policies have affected poverty neighborhoods. The nature of the available information precludes a sys- tematic evaluation of the impact of particular federal antipoverty policies in particular neighborhoods, but some inferences can be drawn about the ways in which the most important federal policies have affected poverty neighborhoods in general, and some guesses can be made about the likely effectiveness of recent proposals. The basic methodology of the chapter is a statistical analysis of the neighborhoods. Changes in individual neighborhoods are discussed, but detailed case studies are not presented. The primary sources of information about these neighborhoods are the decennial censuses of population and housing for 1970 and 1980. (Specifically, data have been taken from the Fourth Count Summary Tape for the 1970 census, and from Summary 68

HOW POVERTY NEIGHBORHOODS ARE C~4NGING 69 Tape File 4 for the 1980 census.) The data for census tracts in slum neighborhoods have been aggregated to provide information on persistent slum areas in big cities. The data were produced as a special tabulation for this study by the Princeton University Computer Center, under a contract with the American Enterprise Institute. Changes in the 1960s are also analyzed, using data for a smaller set of poverty neighborhoods that had been compiled from the decennial censuses of 1960 and 1970, as part of a previous study.) That data set is used to supplement the basic analysis of the 1970s. It permits a somewhat longer perspective on slum neighborhoods, but the information for the 1960s is more limited, in several ways, and it was originally compiled for a different purpose. THE CONCEPT OF A "POVERTY NEIGHBORHOOD" The poverty neighborhoods induded in this study consist of contiguous census tracts, each having 20 percent of its population below the poverty line in both 1970 and 1980. The neighborhoods are delineated with consistent boundaries in both censuses. The 20 percent figure was established after the 1970 census as the cutoff for categorizing a census tract as a low-income area. The same criterion was used in 1980, but the term was changed to poverty area. A different and more complicated concept was used in the 1960s. As part of the War on Poverty, the Census Bureau and the Office of Economic Opportunity developed an index of poverty, which gave equal weight to five factors: family income (unadjusted for family size), children not living with both parents, adults with less than an eighth-grade education, unskilled male workers, and substandard housing. The lowest ranking 25 percent of all census tracts were then classified as poor (Bureau of the Census, 1966~. The 20 percent figure was chosen after the 1970 census as the best approximation to the 1960 method, in terms of classifying tracts as poverty areas (see Bureau of the Census, 1973; Putnam, 1973~. There is a distinction between being poor and living in a poverty neighborhood. Poverty status for individuals depends only on their own income (adjusted for age and family size); poverty status for neighborhoods depends on the incidence of poor individuals within a geographic area. In 1980, for example, 47.5 million people were living in the 100 largest central cities, of whom 8.1 million were poor, 15.5 million lived in poverty areas, and 5.2 million were poor and lived in poverty areas. These figures imply that 2.9 million poor people did not live in poverty areas and that 10.3 iThe data set was assembled for the National Housing Polipy Review in 1973. The changes in housing conditions during the 1960s are summarized in the report of the National Housing Polipy Review (1974), Ch. 6.

70 INNER-CITY POVERTY IN THE UNITED STATES million living in poverty areas were not poor. Most poor people lived in poverty areas, but most people in poverty areas were not poor. Poverty neighborhoods were delineated by the Office of Economic Opportunity after 1970 as groups of contiguous low-income census tracts with 20,000 or more residents in the aggregate. The smallest of these neighborhoods contained at least three census tracts. The 1970 neighbor- hoods are the basis for the poverty neighborhoods in this study, with some adjustments and qualifications. Poverty Neighborhoods as Neighborhoods Local planners and other experts were consulted in establishing the boundaries of the poverty neighborhoods so that they conformed as closely as possible to the local sense of neighborhood. A large group of contiguous poor tracts was often broken into several neighborhoods on the basis of racial, ethnic, or other differences between the smaller areas. Nonetheless, poverty areas often overlap neighborhoods defined on other criteria, and they sometimes include only part of a single neighborhood. Only 1 of the 12 poverty neighborhoods in Chicago, and 1 of the 8 in Los Angeles, are coterminous with neighborhoods as formally defined by local planners and analysts (e.g., the Chicago Community Inventory). Consistent Neighborhood Boundaries Individual census tracts can be classified as poor in one decennial census and nonpoor in the next, or vice versa. In fact, the actual number of poverty census tracts has increased, and the territorial extent of poverty areas in cities has grown substantially between 1960 and 1980, especially in the 1970s. The geographic spread of poverty between 1970 and 1980 occurred partly because of a measurement problem. The Consumer Price Index (CPI), which is used to adjust the poverty line from year to year, overstated the increase in the cost of homeownership during the inflation of the 1970s.2 Even if the error is corrected, however, it seems clear that both the size and the population of poverty neighborhoods in large cities increased markedly. In Chicago, for example, some 227 census tracts, with a total 2 Compared with either the Gross National Product Deflator or the current CPI, the CPI used during the 1970s overstated change in the price level (and therefore the poverty line) By about 6.5 percent, and also overstated the poverty rate in the United States lay about 11.5 percent, or 1.5 percent of the total population, in 1980. The error was corrected, beginning in 1983, but not retroactively. See Weicher (1987) for a fuller discussion of the measurement problem. Corrected data have been published By the Bureau of the Census (1989, Table C and Appendix Err) for yeam beginning in 1974.

HOW POVERTY NEIGHBORHOODS ARE CHANGING 71 population of 818,000, were considered to be poverty areas in 1970. By 1980, the number of tracts increased to 306, with a total population of 1,171,000. The published census reports do not permit the error in the poverty threshold incomes to be corrected on any geographic basis, but it is possible to make a crude adjustment. I estimate that 28 tracts, with a total population of 106,000, would not have been classified as poverty areas in 1980 had the current version of the CPI been in use through the 1970s. (This assumes a rectangular distribution of the population between 75 and 100 percent of the poverty line and uses 92 percent of the official poverty line as the true one.) Even with the correction, there are still 278 tracts and 1,065,000 people in poverty areas in 1980, a larger area and population than in 1970. Some tracts, and even some neighborhoods, have moved out of poverty status. In Washington, D.C., for example, the southern part of the Capitol Hill neighborhood, between the Capitol and Anacostia, was a sizable poverty area (comprising eight tracts) in 1960. It has since largely been gentrified. By 1970, one tract was no longer counted as part of the poverty area; by 1980, four more tracts had been upgraded, and the remaining poverty area consisted of the three southernmost tracts. It seems likely that by 1990 the entire area will disappear from maps of poverty. The neighborhood focus of this study and the growth in the size of poverty areas, taken together, mean that it is not particularly useful to describe changes in poverty areas in the aggregate within a city. Instead, the study focuses on changes in individual poverty areas, keeping the boundaries unchanged. All tracts in each neighborhood were defined as being poor in each of the past three censuses the maximum criterion for "persistence," because the available data only go back to 1960. This maximum criterion was established in order to focus on identifiable neighborhoods that have indeed been persistently poor, persistently slums Watts in Los Angeles, Washington Park in Chicago, the South Bronx in New York. Study Neighborhoods For this study, 79 persistently poor neighborhoods were identified in 12 central cities of standard metropolitan statistical areas (SMSAs). The project has a midwestern and northeastern focus, because these regions have the metropolitan areas with the most persistent slums. There are more midwestern metropolitan areas, but more northeastern neighbor- hoods in the study. This is because New York City is included. Of the 40 Northeastern poverty neighborhoods, 29 are in New York City; in addition, 5 are in Philadelphia, 4 in Washington, D.C., and 2 in Pittsburgh. Six midwestern metropolitan areas are included, with a total of 29 poverty neighborhoods: 12 in Chicago, 5 in Cleveland, 4 in St. Louis, 4 in Kansas

72 INNER-CTTY POVERTY IN THE UNITED STATES City, 3 in Milwaukee, and 1 in Gary. The remaining 10 poverty neigh- borhoods are in the Los Angeles SMSA, 8 from the city of Los Angeles and 2 from Long Beach. The California neighborhoods provide a limited basis for an overview of some regional differences. The sample is obvi- ously dominated by the three supercities of New York, Los Angeles, and Chicago. The differences among these cities constitute an important part of the analysis. Because of the change in the definition of poverty area from 1960 to 1970, and because of occasional redrawing of census tract boundaries, poverty neighborhoods as defined for this analysis need not correspond to the boundaries of poverty neighborhoods as defined in any of the decennial censuses. Few tracts, however, have moved out of poverty between one census and the next. As it happens, 27 of the 79 neighborhoods have the same boundaries in this study as they officially had in 1970. The other 52 have the same boundaries as they officially had in 1960. Their official boundaries in 1980 were larger (often by only one or two tracts) than in 1970; they are examples of the general pattern, previously noted, of increases in the size and population of poverty areas during the decade. The poverty areas defined in this study can be considered the cores of poverty areas as measured by the 1980 census. Redrawing of census tract boundaries limited the number of neighbor- hoods and even eliminated a few large central cities. Boundaries were com- pletely redrawn for the city of Detroit between 1970 and 1980, which makes it impossible to define any poverty neighborhoods consistently. Cincinnati had five poverty neighborhoods in 1960, but only one could have been used in this analysis. The boundaries for so many low-income tracts in Cincinnati were changed that there were only one or two tracts in each of the other neighborhoods that were poor and had consistent boundaries across the three censuses. The loin Cities had only three poverty neighborhoods in 1960, and only two were consistently comparable. Boston had only two poverty neighborhoods in 1960; re-tracting left only one with consistent boundaries by 1980. Buffalo had no neighborhoods with more than To census tracts. The only eastern cities that met the study criteria but were excluded were Baltimore (four poverty neighborhoods), and Newark (three). If the study had been given a national scope, a number of southern and western cities could have been included, but only the San Francisco-Oakland area has as many as five eligible neighborhoods. Houston might also have five or more, but it was completely re-tracted between 1960 and 1970. Most of the larger SMSAs in these regions did not have many poverty neighborhoods in 1960; they developed sizable identifiable concentrations of the poor only as they subsequently grew rapidly. But even with a full set of poverty neighborhoods for all large central cities the 100 largest, for example

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-1 Poverty Neighborhoods in Study and City 73 Number of Population in Poverty Population in Poverty Neighborhoods, Neighborhoods (000), Neighborhoods (000), 1970 1970 1980 City StudyCityStudyCity % Study City % Chicago 12 15 547 818 673941,171 34 Cleveland 5 6 166 207 80103273 38 Gary 1 1 27 33 811564 23 Kansas City 4 4 94 107 8863123 51 Long Beach 2 2 28 57 493385 39 Los Angeles 8 10 367 561 65408964 42 Milwaukee 3 4 76 132 5856161 35 New York 29 35 1,440 2,089 699382,742 34 Philadelphia 5 9 276 494 56197712 28 Pittsburgh 2 3 55 113 4941121 34 St. Louis 4 7 143 284 5088233 38 Washington, D.C. 4 5 106 239 4482251 33 Total 79 101 3,325 5,134 652,4186,900 35 the universe would still be dominated by the largest cities, particularly New York. Limiting the number of metropolitan areas did not greatly reduce the number of neighborhoods. Bible 3-1 presents some basic information about the poverty neighbor- hoods in this study. Most of the neighborhoods officially delineated in 1970 in the cities involved are included, though not all with the same boundaries. The neighborhoods contained well over half of all people living in poverty neighborhoods in these cities in 1970, except for Long Beach, Pittsburgh (both at 49 percent), and Washington, D.C. By 1980, the neighborhoods' total share was down substantially; they had nearly all lost population, and more tracts in the cities were classified as poverty areas. But they still ac- counted for over 30 percent of the people living in poverty areas, except in Gary and Philadelphia. (A list of the tracts contained within each poverty neighborhood, for each of the three decennial censuses, is available from the author.) NEIGHBORHOOD CHAD GES IN THE 1970s Poverty neighborhoods differ from city to city, but the neighborhoods in eastern and midwestern cities changed in similar ways in the course of the 1970s. These changes are more important in understanding the problems of persistent slums than are the surface differences. It is possible, therefore, to summarize the changes in eastern and midwestern neighborhoods in terms of averages for all 69 neighborhoods from the two regions in the

74 INNER-CI~IY POVERTY IN THE UNITED STATES sample (liable 3-2~. This is not to say that all of these neighborhoods fit the pattern shown in the table. There were some differences, by region and by size of city, particularly in housing conditions. But the table is a reasonable representation of what happened in most neighborhoods. The few major differences between cities are discussed in a subsequent section. Eastern and Midwestern Poverty Neighborhoods: A Study in Decline The most important similarity among the neighborhoods in the East and Midwest is also the most important and most obvious regional differ- ence between them and the neighborhoods in Los Angeles. In the East and Midwest, nearly all poverty neighborhoods lost population during the 1970s. In Los Angeles, nearly all of them gained population. Moreover, they gained a very different population; typically, Hispanics replaced blacks. The only eastern neighborhoods with population increases were two in New York City, one in Manhattan and one in Brooklyn. Both also experienced the same raciaVethnic turnover. The typical neighborhood lost about one-third of its total population in the decade, over half its white population, almost one-third of its black population, and almost one-fifth of its Hispanic population. It gained a small number of persons of other races. The averages in 1970 and 1980 mask substantial differences between cities; the white population was concentrated in the smaller midwestern cities, and the Hispanic population in New YorL Most of the trends are nonetheless universal. Not many neighborhoods had a large white population in 1970, but nearly all lost most of whatever they had by 1980. There were small numerical gains from a small base in the number of persons of other races, in most neighborhoods. There was substantial growth in the Hispanic population, however, in a number of New York neighborhoods; in the Midwest, no neighborhoods were predominantly Hispanic in either year, and about half lost Hispanic residents over the decade. The number of households also declined, but by less than the pop- ulation. The size of the typical household therefore also declined, from about 3.1 to 2.8 persons. The "feminization of poverty" is apparent in these poverty neighborhoods. The number of married couples declined by half, but the number of female-headed households remained about the same. About half the households in the typical neighborhood in 1970 consisted of married couples; about half in 1980 were headed by a single woman. At the same time, the population became older, on average; the number and incidence of children under 18 declined, and the incidence of the elderly increased slightly. Economic changes were similarly pronounced. Real median household income (in 1980 dollars) fell by about $1,700. In 1970, dose to half the

HOW POPERlY NEIGHBORHOODS ARE CHANGING TABLE 3-2 Neighborhood Changes in Midwestem and Eastern Cities 1970 Mean 1980 1970 1980 % Mean % % Change People Population43,30029,200 -30.7 White.7,1003,100 19.9 12.2 Blacks28,60019,500 64.1 66.1 Hispanic6,9005,700 14.6 19.4 Other recess800900 1.7 2.4 Over 65 years3,7003,200 9.1 11.1 Under 18 years16,9009,600 38.5 33.0 Households13,80010,400 -23.8 Married couples6,5003,200 47.2 31.1 Female head5,4005,200 38.6 50.6 Male head, no spouse2,0002,000 14.2 18.3 Income and Economy Median household incomeb$9,100$7,400 -18.2 Labor force (16+ years) Percent unemployed3.77.2 Percent not in labor force49.453.9 Percent employed46.738.9 Percent with earningsC81.158.8 Percent with income transfersC46.461.5 Median years of schooling9.210.3 Housing Housing units14,90012,000 -18.8 Occupied housing units13,70010,400 -23.1 Vacancy rate (%)8.413.3 Percent renter-occupied83.982.4 Space Persons per unit3.152.85 Median rooms per unit4.064.15 Percent crowded17.212.1 Quality Percent without complete plumbing5.26.5 Percent without central heat13.614.1 Percent without heat0.20.5 Age Percent 30+ years old73.071.0 Percent built pre-194073.055.4 Percent <10 years old7.57.9 Percent in 1-unit structures14.616.2 Percent in 5+ unit structures59.059.8 Median home valueb$34,000$24,200 -29.6 Median rentb$131$134 +2.5 Non-Hispanic population. ~ 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980. 75

76 INNER-CITY POVERTY IN THE UNITED STATES individuals over 16 were employed; in 1980, only about 40 percent were. The unemployment rate nearly doubled. Transfer payments apparently became a more important source of income, and earnings from work less important. This inference, however, must be tempered by the fact that the information is reported on a different basis in the two years. In 1970, the Census Bureau published data for "families and unrelated individuals," in 1980 for "households," which can consist of two or more unrelated individuals. Despite the problems with comparability, there probably was a real change in the relative importance of transfer payments and earnings from wore The one positive feature in this picture is the increase in years of schooling among the adult population. The improvement was only from the ninth-grade to the tenth-grade level, however, which means the typical adult in these neighborhoods was well behind the rest of the population. Schooling seems to run counter to the other trends; the improvement does not translate into better economic circumstances. The general impression from these changes is that the black middle class, or perhaps more precisely the lower middle class, was moving out of these neighborhoods. People who could afford to leave were doing so. The number of housing units declined, but less rapidly than the number of people or households. This implies that the vacancy rate increased. Vacant units, however, fall into several categories, and the meaning of "vacancy rate" varies among these neighborhoods. Some vacant units are "available for sale or for rent." Others fall into an "other vacant" category, of which the most prominent subcategory is "in boarded-up buildings." These units may be about to drop out of the housing stock altogether. They may be almost indistinguishable from structures that are no longer counted as part of the housing inventory. Boarded-up buildings account for most of the vacant stock in New York and Philadelphia, but a minor fraction of it in the other cities. The age distribution of the housing stock did not change much, but this fact can be misleading. A substantial share of the prewar housing stock- over 4,000 units in the average neighborhood-was razed or otherwise removed from the inventory. The remaining housing aged 10 years, and a small number of new units were built. It is reasonable to conjecture that some if not all of the new units were built under government housing programs for the poor, but this is only a conjecture; unfortunately, the census tract data do not identify public housing or other subsidized housing. Evidence from the 1960s, to be discussed later, suggests that about half the new units added to poverty neighborhoods in that decade were subsidized. There was little change in tenure or structure type. Real home values declined in the course of the decade, but real rents rose slightly. The

HOW POVERTY NEIGHBORHOODS ARE CHANGING 77 rent figure is obviously the more relevant, because few units were owner- occupied. One basic measure of housing conditions improved noticeably over the decade. Households enjoyed more space in 1980. The number of persons per unit declined and the number of rooms per unit increased; fewer units were crowded. The most surprising feature is the increase in the percentage of housing units with fundamental deficiencies. This does not necessarily mean that individual units deteriorated. Because the number of units declined, the increased percentages actually represent stable or declining numbers of units. The average number of units without complete plumbing was almost unchanged, at about 775, and the number without central heat declined from about 2,000 to about 1,700. The change in units without any heat could well reflect sampling error; on average fewer than 100 units per neighborhood reported this problem in each year, and the question was asked only on the census "long form," which was sent to 20 percent of housing units. The change in housing units without complete plumbing facilities is especially noteworthy. The presence of plumbing is the one measure of housing quality reported in every decennial census of housing since the first one in 1940, and the one attribute included in every definition of adequate or standard quality housing since World War II. There has been dramatic improvement in each decade, both for the population as a whole and for the poor and minorities, insofar as data are available. There was dramatic improvement in poverty neighborhoods in the 1960s. After all the progress, the data for these neighborhoods in the 1970s strike a sour note. The increase turns out to be the result of a little-noticed change in definition, combined with a much-noticed change in the housing stock In 1970, a unit was deemed to have complete plumbing if the facilities were located within the structure; in 1980, it was necessary that they be located within the unit. This changes the classification of rooming houses, but that is not the most important effect, statistically. More important is the afore- mentioned increase in vacant, boarded-up buildings, particularly in New York and Philadelphia. The poverty neighborhoods in these two cities were almost the only ones reporting an increase in the incidence of units without plumbing. In the Philadelphia neighborhoods, more than 60 percent of the units lacking complete plumbing in 1980 were in such buildings, compared with less than 20 percent in the Chicago neighborhoods. Intraregional Changes Bibles 3-3 through 3-6 show the same information separately for Chicago, New York, and the smaller cities in each region. (Bible 3-6

78 INNER-CITY POVERTY IN THE UNITED STATES TABLE 3-3 Neighborhood Changes, Chicago 19701980 1970 1980 % MeanMean % % Change People Population45,60032,900 - 25.9 White.5,1002,600 14.6 8.8 Black.37,40026,700 76.2 77.5 Hispanic2,6003,000 8.1 11.8 Other recess500600 1.1 1.9 Over 65 years3,8003,400 8.1 9.7 Under 18 years19,20012,100 42.3 37.6 Households14,20011,200 -20.2 Married couples6,1003,000 43.7 28.0 Female head5,5005,900 38.6 52.6 Male head, no spouse2,5002,300 17.7 19.4 Income and Economy Median household incomeb$9,300$6,900 -26.8 Labor force Percent unemployed4.18.3 Percent not in labor force49.255.7 Percent employed46.636.0 Percent with eamingsC79.054.5 Percent with income transfersC46.165.1 Median years of schooling9.310.2 Housing Housing units15,70012,500 -19.8 Occupied housing units14,10011,200 -19.4 Vacancy rate (%)11.010.6 Percent renter-occupied89.889.1 Space Persons per unit3.263.01 Median rooms per unit3.974.09 Percent crowded19.915.8 Quality Percent without complete plumbing8.56.5 Percent without central heat19.423.3 Percent without heat0.30.3 Age Percent 30+ years old65.265.9 Percent built pre-194065.249.7 Percent <10 years old11.37.9 Percent in 1-unit structures6.46.2 Percent in 5+ unit structures67.568.0 Median home valueb$37,800$28,900 -29.4 Median rentb$155$134 -13.4 Non-Hispanic population. bIn 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980.

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-4 Neighborhood Changes, Other Midwestern Cities 1970 Mean 1980 1970 1980 % Mean % % Change People Population29,70019,000 -34.8 White.7,6003,900 29.0 21.7 Blacks21,60014,300 68.7 73.4 Hispanic700600 2.9 3.6 Other recess200200 0.7 1.6 Over 65 years3,4002,700 11.6 14.2 Under 18 years10,9005,700 36.2 30.2 Households9,9007,100 -27.9 Married couples4,4002,100 44.7 29.7 Female head3,9003,500 38.8 49.2 Male head, no spouse1,6001,500 16.5 21.2 Income and Economy Median household incomeb$9,000$7,900 -11.5 Labor force Percent unemployed4.27.3 Percent not in labor force46.251.1 Percent employed49.541.5 Percent with eamingsC86.563.5 Percent with income transfersC46.261.4 Median years of schooling9.410.6 Housing Housing units11,2008,400 -24.1 Occupied housing units9,8007,100 -27.7 Vacancy rate (%)11.715.8 Percent renter-occupied72.869.1 Space Persons per unit3.022.70 Median rooms per unit4.224.42 Percent crowded12.27.5 Quality Percent without complete plumbing6.64.7 Percent without central heat21.617.4 Percent without heat0.20.3 Age Percent 30+ years old77.876.0 Percent built pre-194077.860.0 Percent <10 years old5.56.1 Percent in 1-unit structures23.930.1 Percent in 5+ unit structures36.435.8 Median home valueb$26,000$16,800 -34.4 Median rentb$116$105 -9.6 Non-Hispanic population. bIn 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980. 79

80 INNER-C~ POW IN THE UNFED STATES TABLE 3-5 Neighborhood Changes, New York 1970 Mean 1980 1970 1980 % Mean % % Change People Population49,70032,400 -31.7 White.7,9002,500 18.7 7.9 Blacks26,30017,300 50.5 52.5 Hispanic14,00011,000 27.8 36.0 Other races.1,5001,600 3.0 3.6 Over 65 years3,6003,000 7.3 8.8 Under 18 years19,80010,800 39.9 34.5 Households15,70011,600 -23.6 Married couples7,9003,900 52.0 34.4 Female head5,9005,700 37.4 50.2 Male head, no spouse1,8002,000 10.6 15.4 Income and Economy Median household incomeb$9,400$7,300 -22.8 Labor force Percent unemployed3.16.4 Percent not in labor force51.654.2 Percent employed45.039.3 Percent with earningsC77.158.2 Percent with income transfersC45.058.4 Median years of schooling9.110.0 Housing Housing units16,20013,100 -16.9 Occupied housing units15,50011,600 -22.7 Vacancy rate (%)4.711.6 Percent renter-occupied91.791.7 Space Persons per unit3.242.87 Median rooms per unit3.943.90 Percent crowded20.614.2 Quality Percent without complete plumbing3.37.2 Percent without central heat4.66.7 Percent without heat0.20.5 Age Percent 30+ years old71.367.7 Percent built pre-194071.352.5 Percent <10 years old7.39.5 Percent in 1-unit structures3.32.7 Percent in 5+ unit structures77.578.8 Median home valueb$44,500$30,200 -31.2 Median rentb$133$155 +17.4 Non-Hispanic population. bIn 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980.

[ION POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-6 Neighborhood Changes, Other Eastern Cities 19701980 1970 1980 % MeanMean % % Change People Population45,30033,300 -26.3 Whites6,2004,000 13.6 12.2 Blacks37,30026,700 82.8 81.0 Hispanic1,5002,100 2.7 5.6 Over races'400400 0.9 1.3 Over 65 years4,8004,400 11.3 13.9 Under 18 years16,0009,400 33.7 27.1 Households14,90011,800 -21.0 Married couples6,1003,200 40.8 27.4 Female head6,3006,100 42.2 51.8 Male head, no spouse2,5002,500 17.0 20.8 Income and Economy Median household incomeb Labor force Percent unemployed Percent not in labor force Percent employed Percent with earningsC Percent with income transfersC Median years of schooling $8,800 $8,400 4.1 7.7 48.6 47.1 86.7 51.6 9.4 55.6 36.5 57.8 67.2 10.6 -2.3 Housing Housing units16,700 14,500 -13.7 Occupied housing units14,800 11,800 -20.8 Vacancy rate (Jo)10.9 17.9 Percent rented71.7 67.5 Space Persons per unit2.99 2.80 Median rooms per unit4.23 4.53 Percent crowded12.4 8.9 Quality Percent without complete plumbing3.9 6.5 Percent without central heat19.5 19.8 Percent without heat0.3 1.2 Age Percent 30+ years old79.7 78.8 Percent built pre-194079.7 63.8 Percent < 10 years old6.9 6.1 Percent in 1-unit structures44.5 47.0 Percent in 5+ unit structures31.2 32.9 Median home valueb$27,300 $30,300 +6.1  Median rentb$128 $124 -2.6 Non-Hispanic population. ban 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980. 81

82 INNER-CITY POVERTY IN THE UNITED STATES is included mainly for the sake of completeness; it presents averages for Philadelphia, Washington, and Pittsburgh, three quite different cities with different experiences during the 1970s.) Some of the differences in popu- lation and housing characteristics in 1970 and 1980 are very noticeable in these tables, but they are less important than the similarities in trends. The housing stock in New York neighborhoods, to take an obvious example, consisted much more of large apartment buildings in both 1970 and 1980, and surely does today, than in neighborhoods in any other city. In New York and Chicago, rental housing was much more predominant than in the smaller cities; the typical housing unit was smaller, and crowding (de- fined as having more than one person per room) was much more common. Rents and values were higher. There were also some demographic differ- ences. New York had a much higher Hispanic population, and midwestern cities (except Chicago) had many more white persons living in poverty neighborhoods. Most of the demographic and economic trends, by contrast, are similar. The most noticeable demographic difference is the growth in the Hispanic population in New York neighborhoods. This was concentrated in a few neighborhoods, but 25 of 29 neighborhoods experienced a percentage in- crease in the Hispanic share of the population. In addition, real income trends differed. Household income declined more in the two big cities, and hardly at all in smaller eastern cities. This latter result occurs because income increases in poverty neighborhoods in Washington offset declines in Philadelphia and Pittsburgh. Some other differences in the changes are statistically significant, but do not matter qualitatively. Housing trends were more diffuse. The vacancy rate went down in Chicago but rose elsewhere. The average for Chicago was driven by substantial declines in the Cabrini-Green public housing complex and the Woodlawn neighborhood; most other neighborhoods experienced small increases. The overall increase in median rents, shown in Table 3-2, is the net effect of regional divergences: an increase in New York and Washington, a decrease elsewhere. The data for home values in the other eastern cities similarly hide a diversity within the region: values went up by at least 35 percent in each of the four Washington neighborhoods, and down by at least 30 percent in each of the seven Philadelphia and Pittsburgh areas. The apparent inconsistency in the data on mean home value for the smaller eastern cities occurs because the percentage change is calculated as the mean of the percentage changes for the individual areas, rather than as the percentage change of the means from 1970 to 1980. If the range of decennial mean values is large, the percentage change of the mean is likely to differ from the mean of the percentage changes. A similar phenomenon occurs in some other tables.

HOW POVERTY NEIGHBORHOODS ARE CHANGING 83 The overall increase in the percentage of units without central heat results partly from the phenomenon of boarded-up buildings in New York and Philadelphia, and partly because many of the high-rise public housing projects built in Chicago apparently do not have central heating systems. The highest incidence of units without central heat-over two-thirds- occurs in the neighborhood containing the Cabrini-Green project. Over 75 percent of the residents in the tracts containing the project reported that they did not have central heating. Census tracts in other areas with a signif- icant concentration of public housing also reported a substantial incidence of units without central heating. As the older private housing stock was removed from the inventory during the 1970s, public housing constituted a larger share of the inventory, and the proportion of units without central heat increased. Thus, the reported rise in the Chicago neighborhoods- certainly unexpected at first sight, considering the climate-does not by itself indicate a deterioration in housing quality. Los Angeles Los Angeles is fundamentally different from the midwestern and east- ern cities. All but two of its poverty neighborhoods gained population, in every case because the growth in the Hispanic population more than offset declines in whites and blacks (Table 3-7~. The incidence of the el- derly declined, and the incidence of children increased. The number of households remained about the same, and the average number of people in each household increased. There was no shift away from married couples to female-headed households; both increased slightly, and the share of households headed by a single man declined. Changes in the labor force were also minor, compared with those in the East and Midwest. There was a small increase rather than a large decrease in labor force participation, and the unemployment rate went up much less. There may also have been a smaller shift away from earnings from work, and toward transfers, as sources of income, though the change in the data makes this uncertain. It is clear, however, that in 1980 more households relied on earnings from work, rather than transfers, for their income, again unlike the East and Midwest. Real household income went up slightly, rather than down; and years of schooling remained about the same, rather than rising. Housing conditions were consistent with the growth in population. The housing stock was utilized more intensively. There was a small decline in the total number of units but no change in the number of occupied units. The vacancy rate fell. The average number of people per unit increased while the number of rooms per unit declined, resulting in a sharp increase in crowding. Quality changes were ambiguous: Increases in the proportions

84 INNER-C17Y POVERTY IN THE UNITED STATES TABLE 3-7 Neighborhood Changes, Los Angeles and Long Beach 1970 Mean 1980 1970 1980 % Mean % % Change People Population 39,500 44,100 +13.5 White. 6,500 4,900 22.7 13.6 Black. 20,500 16,300 44.6 35.8 Hispanic 10,500 20,300 27.3 44.6 Other races. 2,000 2,600 5.3 6.1 Over 65 years 4,300 4,200 12.0 9.8 Under 18 years 13,600 14,400 33.9 33.8 Households 14,300 14,100 -0.1 Married couples 5,800 6,300 42.8 43.6 Female head 5,100 5,400 36.0 38.9 Male head, no spouse 3,300 2,400 21.4 17.5 Income and Economy Median household incomeb $8,200 $8,500 +4.2 Labor force Percent unemployed 5.3 6.3 Percent not in labor force 50.3 48.3 Percent employed 43.6 45.3 Percent with eamingsC 79.7 66.2 Percent with income transfersC 52.2 53.2 Median years of schooling 10.0 10.0 Housing Housing units 15,400 15,100 Occupied housing units 14,200 14,100 Vacancy rate (%) Percent renter-occupied Space Person per unit Median room per unit Percent crowded Quality Percent without complete plumbing Percent without central heat Percent without heat Age Percent 30+ years old Percent built pre-1940 Percent <10 years old Percent in 1-unit structures Percent in 5+ unit structures Median home valueb Median rentb 7.5 78.0 2.85 3.40 17.7 S.4 75.6 4.8 54.0 54.0 11.0  41.7 39.4 $43,700 $128 6.3 79.1 3.17 3.24 26.2 6.5 62.3 7.5 60.6 38.8 8.2 39.5 43.6 $48,400 $152 -0.6 +0.6 +9.5 +18.6 Non-Hispanic population. bIn 1980 dollars. CFor families and unrelated individuals in 1970, households in 1980.

HOW POVERTY NEIGHBORHOODS ARE CHANGING 85 of units without complete plumbing and units with no heat were offset by a larger decrease in units without central heating. Heating equipment is not a basic indicator of housing quality in Los Angeles, so perhaps only the change in units without complete plumbing merits attention. The housing stock was newer in Los Angeles, and it aged a bit more in the course of the decade, although there was a decline of about one-third in the number of prewar units. Both rents and values increased, reflecting the greater demand for housing in these areas and the population growth in Los Angeles as a whole. The decline in black population in these neighborhoods deserves spe- cial attention. It mainly indicates a geographic dispersal of the black population from these poverty neighborhoods, rather than any dramatic improvement in the status of blacks. There was virtually no change in the various measures of poverty and location from 1970 to 1980. The total number of blacks in the city, the number who were poor, the number living in poverty areas, and the number of poor blacks in poverty areas-all changed by less than 5,000 in 10 years. The number of census tracts classi- fied as poverty areas, however, increased from 162 to 234. Most of the Los Angeles poverty neighborhoods in this study are close to the Civic Center. The poor black population in these neighborhoods apparently moved to nearby census tracts, causing many of them to be classified as poverty areas in 1980 and causing a spread of poverty to a wider area around the Civic Center, particularly to the southwest and south. NEIGHBORHOOD CHANGES IN A LONGER PERSPECTIVE Although comparable data for the same set of neighborhoods could not be obtained from the 1960 census, as noted, data were available for a different set. These areas were chosen on the same principles: They were poverty areas, or the cores of poverty areas, in both 1960 and 1970. They were chosen for a different purpose, however, namely to analyze changes in housing conditions as part of a national housing polity study. The information about these neighborhoods is therefore more limited; it includes most of the housing data, but otherwise only population and income. Household composition and economic status were omitted. The neighborhoods were also selected on a different basis. The data set includes just a few neighborhoods, typically one or two, from each of 29 cities, and all parts of the country are represented. The total number of neighborhoods is only 50 (Bible 3-8~. Seven neighborhoods turned up in both samples, which makes it pos- sible to look at the changes in these neighborhoods on a consistent basis over both decades. (Another 12 neighborhoods differed by only one or two census tracts, but a longitudinal analysis of these areas was not attempted.

86 INNER-CITY POVERTY IN THE UNTTED STATES TABLE 3-8 LOcahon of Poverty Neighborhoods in 1960-1970 Data Set Region City Number of Neighborhoods East Midwest South West New York Baltimore Boston Philadelphia Washington, D.C. Newark Pittsburgh Chicago St Louis Cincinnati Cleveland Detroit Gary Indianapolis Milwaukee Minneapolis Atlanta Dallas Houston Birmingham Memphis Miami New Orleans San Antonio  Los Angeles Denver Oakland San Diego San Francisco 14 4 2 2 2 2 1 1 18 2 2 2 11 2 2 2 1 In every instance, the boundaries were larger in the later data set, and the added tracts were higher income areas, with better housing; a longitudinal analysis would therefore understate the decline in population and overstate the improvement in income and housing conditions.) Population and Housing Changes in the 1960s Table 3-9 summarizes the population and housing conditions in 1960 and 1970 for the sample of 50 neighborhoods. All but one lost population

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-9 Neighborhood Changes in the 1960s 1960 Mean 1970 1960 1970 % Mean % % Change Population and locome Population54,20043,200 -20.1 White'22,80012,900 42.0 29.9 Blackb28,00026,000 51.7 60.1 Hispanic3,4004,300 6.3 10.0 Median household incomes$4,500$4,700 +4.1 Housing Housing units18,80016,100 -14.3 Occupied housing units17,20014,500 -15.5 Vacancy rate (%)8.611.5 Percent rented81.085.0 Space Persons per unit3.162.68 Median rooms per unit3.773.95 Percent crowded18.415.1 Percent without complete plumbing19.710.2 Percent in 1-unit strictures36.241.7 Median home values$14,500$13,300 -9.5 Median rents$77$76 -1.3 aExcludes Hispanic population, includes persons of other races. Non-Hispanic population. can 1970 dollars. 87 over the decade; the exception was a neighborhood in Miami that received an influx of Cuban refugees. Coincidentally, this data set includes two of the Los Angeles neighborhoods that had a growing Hispanic population in the 1970s; both were much less Hispanic in 1960 and lost population during the decade. They did, however, lose less population than the average during the decade, as was also true of the few other neighborhoods with a substantial Hispanic minority in 1960. The housing changes in the Miami neighborhood in the 1960s were much like the changes in the Hispanic neighborhoods in Los Angeles in the 197Qs: a lower vacancy rate, higher rents, more crowding, and an increase in units without complete plumbing. As far as the data go, they suggest that Hispanic poverty neighborhoods were consistently different from white Anglo and black poverty areas. In general, the data for the 1960s are much like the data for eastern and midwestern neighborhoods in the 1970s. The average decline in popu- lation was smaller in the 1960s about one-fifth versus one-third-but this can be accounted for by the small declines in the southern and western neighborhoods that were excluded from the later data set. As in the 197Os, there was a 50 percent decline in the white population (including other races), and a modest gain on average in the number and incidence of

88 INNER-C1IY POVERTY IN THE UNITED STATES Hispanics. The black population declined on average, but much less than in the 1970s. It appears that the white middle class may have been moving out during the 1960s, the black middle class in the 197Os. The economic circumstances of the people in these neighborhoods improved in the 1960s. There was a modest gain in real income, unlike in the 1970s. The difference may be explained by the fact that the earlier decade saw a much greater rate of economic growth for the country as a whole. Housing changes were similar in both decades. There was a decline in the number of units, a larger decline in the number of occupied units, and a rise in the vacancy rate. The typical household had more space: There were fewer persons per unit, more rooms per unit, and a lower incidence of crowding. For the small number of homeowners, values declined in real terms; for the large number of renters, real rents were stable. Data on the age of the housing stock were not collected on a comparable basis in 1960, but it seems likely that the experience was similar-a large loss among the oldest housing units and a small volume of new construction. The big difference between the decades is the big drop in units without complete plumbing in the 1960s. This is the main reason why the reported increase in the 1970s was so surprising; it was inconsistent with the previous decade. From 1960 to 1980 The seven neighborhoods in both data sets confirm the consistency of the long-term trends. Data for these neighborhoods are shown in liable 3-10. Many of the neighborhoods have been nationally known for their housing or other problems: Bedford-Stuyvesant in Brooklyn, Watts in Los Angeles, Hough in Cleveland, Woodlawn and East Garfield Park in Chicago, as well as neighborhoods in Philadelphia and St. Louis. What has happened in them, as their notoriety has faded, may be of interest for its own sake, as well as being potentially indicative of changes elsewhere. The seven neighborhoods lost about half their population from 1960 to 1980. Nearly all the white population moved out in the 1960s, and half of the small remainder moved out during the next decade. The black population declined slightly in the 1960s and substantially in the 1970s; its share grew in the first decade and was stable in the second. None of these neighborhoods had a large Hispanic or other population at any time; they are perhaps representative of the older slums in the older cities. The typical family enjoyed a modest increase in real income over the 1960s, followed by a 25 percent decrease in the 1970s; the net effect was a substantial decline. In each decade, households in these areas fared much worse than the typical American household.

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-10 Average Neighborhood Changes, 1960-1980 89 1960 1970 1980 Population63,70048,30032,800 Whitea15,5004,1002,000 Blacka45,00041,00027,300 Hispanic3,1003,0003,300 Other races100200200 Median household incomes$9,500$10,400$7,900 Housing units Occupied Vacancy rate (%) Percent single family Percent owner occupied Persons per unit Rooms per unit Percent crowded Percent without complete plumbing Median valueb Median rentb 19,200 18,100 6.0 36.0 17.4 3.64 4.11 23.0 16.2 $45,500$33,300$22,400 $151$139$129 16,000 14,500 9.1 30.1 26.6 3.35 4.44 17.3 4.8 12,900 1 1,000 13.1 32.2 29.2 3.07 4.56 13.0 5.5 On-Hispanic only. 1980 dollars. SOURCES: Special tabulations of 1960 census tracts by Oak Ridge National Laboratory, and of 1970 and 1980 tracts by Princeton University Computer Center. The housing stock decreased, the vacancy rate increased, and real  rents fell over both decades. All measures of housing space showed steady improvement. The plumbing data for 1980 are affected by the boarded- up buildings in New York and Philadelphia; the other five neighborhoods showed a decline in units without complete plumbing in both decades. The data on homeownership and home values can be misleading. The neighborhoods in Los Angeles and Philadelphia have predominantly single- family housing; the St. Louis neighborhood was about half single-family in 1960. The Philadelphia neighborhood accounted for more than half the total number of owner-occupied units; the Los Angeles and St. Louis neighborhoods, nearly all the rest. Average changes therefore should be viewed with caution. In the 1960s, only Watts experienced an increase in real home values, of about 10 percent. All the others, including Philadelphia, had a nominal increase of less than 10 percent, or a slight nominal decrease. The 1970s saw a similar regional disparity. The average changes shown in the table slightly understate the declines in the East and Midwest-declines that really matter only for Philadelphia and St. Louis and completely misrepresent what happened in Watts.

go INNER-CITY POVERTY IN THE UNITED STATES THE FILTERING PROCESS AND THE CULTURE OF POVERTY The Bottom of the Barrel The changes in the eastern and midwestern poverty neighborhoods suggest that a "filtering process" is at work in the metropolitan areas as a whole, and these neighborhoods are the dregs. The term "filtering" has been used to describe many processes of change in urban areas. In this study, it is used to describe a system of changes in residential location and changes in neighborhood housing costs. The filtering process begins when economic growth makes it possible for high-income families to move into new housing. As they move out, their old housing becomes available and more affordable to middle-income families, who then are able to move to the suburbs that the high-income families are leaving. Their housing in turn becomes available to families with somewhat lower incomes. Eventually, some poor people living in poverty neighborhoods are able to move into housing outside the neighborhood, housing that moderate-income families are leaving. These poor people are likely to be relatively the richest and most stable, economically and socially. They leave behind the poorest of their neighbors. As the poverty neighborhoods empty out, the worst housing within them drops out of the housing stock or stands vacant. The smallest and least desirable units are often but not always the oldest. The decline in population reflects a decline in the desirability of the neighborhood, so rents and values fall even though the housing stock, on average, has improved. -o r Mobility Between and Within Neighborhoods The "filtering" metaphor has some limitations, however. The neigh- borhoods do not "empty out" so much as they "churn." Mobility data for all poverty neighborhoods in these cities (not limited to the persistent slums) show that the overwhelming majority of residents in 1980 moved into their apartments after 1970. Between 45 and 62 percent had moved in since 1975; between 63 and 78 percent during the full decade. These mobility rates were typically just a few percentage points higher than those for the remainder of the cities, so not too much should be read into them. Nonetheless, they may have policy implications for community development or neighborhood revitalization efforts. From that perspective, the question of inter- vs. intra-neighborhood mobility may be important. If most residents have moved within the neighborhood, or for a short distance, then the population at the end of the decade includes many people who were there at the beginning, and the stability of the population might give the neighborhood some basis for revitalization. If instead most of the

HOW POVERTY NEIGHBORHOODS ARE CHANGING 91 residents come from some other neighborhood, and few stay for more than a few years, then any effort at revitalization is likely to be much more difficult, because the population may have very little sense of neighborhood attachment. The census data identify previous residence only by municipality, not by neighborhood. Most of the residents of these neighborhoods moved in from elsewhere inside the city, if they did not live in the same house for the full decade. There have been a few studies of neighborhood mobility for the poor, for example as part of the Supply Experiment within the Experimental Housing Allowance Program. They show that most poor people move within the neighborhood, or for a short distance (see for example, Butler et al., 1969; Greenberg and Boswell, 1972; Lowry, 1983~. This limited evidence is not fully relevant, however, because poor people, especially renters, move frequently. If they move three or four times in the course of a decade, one of those moves might be to a different neighborhood, even if all the others covered a short distance. Some of the evidence in this study is consistent with a pattern of predominantly intra-neighborhood moves. The richest households and the worst housing units are simultaneously "leaving the neighborhood," so the remaining poor people may be moving to the better housing. But this argument cannot be pushed too far. It implies that there would be a greater improvement in housing quality in neighborhoods that lost the most people. In fact, there was no correlation between population change and quality improvements, except for crowding. The argument also implies that, for neighborhoods with larger population declines and neighborhoods with more improvement in housing quality, fewer of the residents would be living in the same house at the end of the decade. These correlations also do not appear in the data. It is possible that the neighborhoods, bad as they are, could be attract- ing low-income households from other areas because the local housing is a bargain. It appears that real rents for housing of a given quality declined over the decade, although this cannot be demonstrated rigorously. If that did occur, low-income people could have moved in to take advantage of the fall in price. A more detailed inspection of some Chicago census tracts reveals an interesting pattern. Turnover is much lower in public housing projects than in private low-income housing. In Cabrini-Green, 59 percent of the households had lived in the same unit for five years; in the rest of that poverty neighborhood, the proportion was 40 percent. Similar but less pronounced differences occur in other areas, for instance between Henry Homer Homes and the rest of the Near West Side, and between the high- rise projects stretching south of the Loop along State Street and the nearby older, low-rise private housing.

92 INNER-CII~Y POVERTY IN THE UNITED STATES The New Immigration The poverty areas in Los Angeles are more like the traditional slums of the older big cities, at least in a statistical sense, than the old neigh- borhoods are themselves. They are ports of entry into the United States, or into the urban United States. About 10 percent of the 1980 residents of these areas had moved into the United States since 1975, presumably from Mexico and Latin America. Virtually nobody moved into the Chicago poverty neighborhoods from abroad. Population densities in the Los An- geles neighborhoods are lower than in the neighborhoods in the eastern and midwestern cities, but in the local context they are "teeming slums." Indeed, the housing is much more crowded in Los Angeles than in the East, though the density of dwelling units per acre is clearly lower. They are also vibrant rather than stagnant. People are working and doing better, rather than living on transfer payments and doing worse. STATISTICAL ANALYSES OF NEIGHBORHOOD CHANGES Support for the notion of a filtering process at work comes from more sophisticated statistical analyses of some of the changes in the 79 poverty neighborhoods. The dependent variables include population, income, un- employment, and several measures of housing conditions. In each case, changes in the poverty neighborhoods are related both to changes in the metropolitan areas as a whole and to other changes in the neighborhoods that might be causal factors. Separate regression results are reported in the tables below for all 79 poverty neighborhoods, all neighborhoods except Los Angeles, and all neighborhoods except Los Angeles and New York. Los Angeles is omitted because of the very different changes that occurred there, New York because the large number of poor neighborhoods could dominate the analysis and obscure important relationships in other cities. The variables in the regressions are generally expressed in percentage terms, either as percentage changes over the decade in population, income, or the stock of housing, or as changes in the percentage of units falling into a given category, such as the change in the percentage of the population that is Hispanic, or the change in the percentage of households that are headed by a woman. The reported regressions usually include only statistically significant variables; some interesting insignificant results-dogs that did not bark are discussed in the text. General Findings Before turning to the detailed discussion of relationships, it is worth- while to take a broader perspective. The 79 neighborhoods are clearly

HOW POVERTY NEIGHBORHOODS ARE CHANGING 93 integral parts of their metropolitan areas, not isolated islands. Population, income, and unemployment changes in the neighborhoods are all signifi- cantly and positively related to the corresponding changes in the SASH If the SMSA prospered, the poverty neighborhoods shared in the prosper- ity. The relationships are not proportionate, however. Neighborhood gains were smaller than metropolitan gains for population and income, but larger for the unemployment rate. There is also statistical evidence of filtering. The concept is measured as the rate of new housing production relative to net household formation during the 1970s, for the SMSA as a whole. A high ratio-implying a greater possibility for filtering is associated with better neighborhood housing conditions and perhaps with a greater decline in neighborhood population. The most consistently important neighborhood factor is the growth in Hispanic population, which tends to put pressure on the local housing stock at the same time that local income and employment rise. Changes in household composition also matter. New Housing and Neighborhood Changes Else effects of new housing on the neighborhood deserve special men- tion. Building low-income housing in slum neighborhoods has been a public policy since the 1930s. Census data do not identify subsidized housing; new housing in the aggregate is the nearest approximation to new subsidized housing, though it is not the same thing. Even in persistent slums, some housing construction is reported in the decennial census. The data for the 50 poverty neighborhoods in the 1960s included new subsidized units from a special tabulation produced for the U.S. Department of Housing and Urban Development (HUD), showing the location of housing projects by census tract.3 Surprisingly, the HUD programs accounted for only about half of the new units reported in the 1970 census. In several neighborhoods there were no new subsidized units, according to HUD, but 10 to 20 percent of the stock was added during the decade, according to the census. Public and private construction are likely to arise from different cir- cumstances and to have different effects on a neighborhood. New private construction is a sign of neighborhood strength or revitalization; it probably occurs if income is rising and other economic and social characteristics are improving. Subsidized housing, on the other hand, is often built in dete- riorating neighborhoods. New subsidized projects will hold population in 31he data on subsidized housing in the HUD tabulation give the HUD program, the number of units, the census tract, and the year built, through 1972.

94 INNER-CITY POVERTY IN THE UNITED STATES the neighborhood or bring residents in from other areas, but income and employment will not necessarily rise; indeed, they may fall. The indirect evidence is mixed, but it suggests that most of the new housing in these neighborhoods is subsidized. Housing construction during the 1970s is consistently and positively associated with population change; neighborhoods decline more slowly. It is not related, however, to most of the measures of housing quality. Construction does result in a decline in crowding in the smallest sample, but the effect is small. There is also a weaker and smaller relationship in the largest sample, significant only at the 20 percent level. In the plumbing and heating regressions, there is no effect at all. This is odd. All new housing units must have complete plumbing, and most will have central heating (perhaps excluding some public housing projects). They must be better, on average, than the units they are replacing, but the improvement does not appear statistically. The explanation may be partly the change in the census definition of "lacking complete plumbing" and partly the fact that not all new housing in these neighborhoods is subsidized by the government. In an earlier analysis of poverty neighborhoods during the 1960s (Weicher, 1973), new subsidized housing had a larger and more significant effect on the decline in units without complete plumbing, than did new private construction. Private construction may be more common in neighborhoods where the rest of the housing stock is relatively good. It is also possible that there are errors in the census data on "year structure built." The information is provided by the resident on the census question- naire; if the current resident is not the original occupant, he or she may not know the correct year. Population Population changes distinguish improving poverty neighborhoods from deteriorating ones, so an explanation of population change is central to the analysis. It is therefore unfortunate even though understandable that the results vary substantially among the samples (Table 3-11~. The high growth rate in Los Angeles, for both total and Hispanic population, results in a growing population for most of its poverty neigh- borhoods. The Los Angeles experience is so different from the other cities that the relationships are statistically significant even though Los Angeles accounts for only 10 of the 79 neighborhoods. When the sample is restricted to the eastern and midwestern neigh- borhoods, New York accounts for almost half of it. For this half, SMSA population changes, and other changes, were the same. At the same time, population changes among New York neighborhoods were more variable

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-11 Regression Analysis of Neighborhood Population Changes 95 Sample Excluding A1179 Excluding Los Angeles Vanable Neighborhoods Los Angeles and New York Constant-.31-.02 -.33 SMSA population change+.93 (3.03 SMSA Hispanic population change+2.09 (2.4) SMSA Filtenng-_ -.59 (1.8) Hispanic population change+.47+.52 +.76 (2.3)(2.4) (3.8) New housing+.61+.55 +.96 (3.1)(2.5) (3.9) Crowded housing in 1970-.67-.83 (2.5)(2~9) One-room units in 1970+.58 (2.8) Old housing units in 1970 -.29 -.71 (2.2) (2.1) R2.68.26 .50 NOTE: Figures in parentheses are t-raiios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentages or percentage changes. than in the rest of the sample. New York had the only two neighbor- hoods with population increases, and also the three neighborhoods with the largest decreases (over 75 percent decline in a decade). None of this variability can be explained statistically by anything that happened to the New York metropolitan area as a whole; all of it has to be explained by changes within the individual neighborhoods. Thus, it is not surprising that no SMSA changes are significant in the East/Midwest regression, and the ability of any regression model to explain population changes is relatively weak. Finally, when New York is excluded, population changes for the re- maining 40 neighborhoods (spread across nine cities) are again affected by changes in the SMS~ The availability of housing in the SMSA the filtering variable apparently makes it possible for people to leave the neighborhood. This relationship is on the margin of statistical significance, but it is reasonable. Neighborhood-specific factors also affect population change. I\vo are significant for all three samples, and also for the New York neighborhoods considered by themselves (not shown in Able 3-11~. One is new housing

96 INNER-CITY POVERTY IN THE UNITED STATES production; the other is the growth in the neighborhood's Hispanic popu- lation. This matters even when the sample is restricted to the cities with small Hispanic populations. There is also some evidence of a "flight from blight," especially in New York. Population declined more in neighborhoods with a higher percentage of old housing units at the beginning of the decade in the full East/Midwest sample, but not when Los Angeles was included or New York was excluded. For the two larger samples, population declined more in the neighborhoods that were the most crowded in 1970. Finally, and oddly, in the full sample, population growth was greater in neighborhoods with a larger share of single-room units rooming houses in 1970. This variable was originally included in the expectation that rooming houses and other small units were becoming increasingly undesirable. The opposite effect appears in the results. This apparently occurs because two Los Angeles neighborhoods with very high concentrations of rooming houses gained population. Income Most of the other changes in these neighborhoods can be explained by the same factors in all of the samples. The decline in real household income, one of the most discouraging trends, is consistently related to both the strength of the metropolitan area economy, and to the change in the incidence of female-headed families (Table 3-12~. Poverty neighborhood incomes tended to rise, or at least to fall less, in the most prosperous metropolitan areas. This indicates that the neighborhoods did not function as isolated economic units apart from the larger regions. At the same time, however, these areas lagged behind the rest of the SMS~ If the metropolitan area enjoyed a 10 percent growth in real household income, the typical poverty neighborhood gained only 4 percent; if the area enjoyed a 5 percent growth, the poverty neighborhood lost about 2 percent. Real income declined the most in the neighborhoods with the greatest percentage increase in the incidence of female-headed households. In the two larger samples, income changes also were positively correlated with changes in the proportion of adults who were employed; the more peo- ple who worked, the better off the neighborhood. In addition, income rose more in the neighborhoods where the housing stock increased. These were mainly the Hispanic neighborhoods, but statistically, however, em- ployment and housing stock changes mattered more than the change in the Hispanic population. When New York and Los Angeles were excluded, neither employment nor housing stock changes were significant. Instead, income change was negatively related to the change in the black popula- tion; incomes fell more in the neighborhoods where the black population increased.

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-12 Regression Analysis of Neighborhood Income Changes 97 Sample Excluding A1179 Excluding Los Angeles Vanable Neighborhoods Los Angeles and New York Constant -.09 -.06 -.06 SMSA income charge +1.28 +1.27 +1.67 (5.5) (5~1) (3.1) Black population change -.58 (2.0) Female-headed household change -.39 -.45 -.84 (5.2) (5.0) (5.2) Employment change +1.03 +1.16 +.65 (4~9) (4~7) (1~3) Occupied housing stock change +.14 +.16 (2.0) (2.1) R2 .67 .59 .66 NOTE: Figures in parentheses are t-raiios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentages or percentage changes. Housing Conditions Three aspects of neighborhood housing are analyzed: crowding, the absence of complete plumbing, and the absence of central heating. In each case, the neighborhood housing market turns out to be greatly affected by the metropolitan market. The change in crowding is affected by population pressures both within and outside the neighborhood: the overall growth rate in the two larger samples, and the neighborhood growth in both total and Hispanic popula- tion in all three samples (Table 3-13~. It also may be related to new housing construction within the neighborhood. The importance of the change in definition (from "within structure" to "within unit") shows up clearly in the analysis of units lacking complete plumbing. The most significant and important factors are the change in the local vacancy rate, which reflects the increase in the boarded-up buildings in New York and Philadelphia, and the change in one-room units, many of them presumably located in rooming houses (Table 3-14~. Conditions in the metropolitan area were also important. A high construction rate allowed units without complete plumbing to drop out of the stock Outside of New York and Los Angeles, also, more units without plumbing dropped out of the stock as more people left the neighborhood. They were probably the worst units to begin with. Metropolitan conditions also affected the incidence of units without central heating (Table 3-15~. A high rate of SMSA population growth put

98 INNER-C~ BERM IN THE UNWED STATES TABLE 3-13 Regression Analysis of Neighborhood Changes in Crowding Sample Excluding Al179 Excluding Los Angeles Variable Neighborhoods Laos Angeles and New York Constant -.00 -.02 +.01 SMSA income change +.30 +.24 (3.6) (2.3) Population change +.14 +.09 +.14 (4~5) (3.5) (2~1) Hispanic population change +.14 +.10 +.27 (2.4) (2.1) (2~9) New housing -.08 -.29 (1.3) (2.0) R2 68 .29 .48 NOTES: Figures in parentheses are t-ratios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentages or percentage changes. TABLE 3-14 Regression Analysis of Change in Housing Units Without Complete Plumbing Sample Excluding A1179 Excluding Los Angeles Variable Neighborhoods Los Angeles and New York Constant+.02+.02 +.04 SMSA filtering-.32-.31 -.24 (3.0)(2~9) (1.6) Vacancy rate change+.27+.20 +.31 (4.8)(2~9) (3~3) Change in one-room housing units+.63+.83 +.82 (6.5)(6.2) (4.2) Change in occupied housing stock +.12 (2.2) R2.55.58 .52 NOTES: Figures in parentheses are t-ratios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentages or percentage changes. pressure on the neighborhood housing stock and kept those units in use. Conversely, a high rate of economic growth and a high rate of housing construction permitted abandonment or destruction, or alternatively up- grading. Outside of New York at least, units without central heating were disproportionately the units vacated or removed from the neighborhood stock.

HOW POVERIY NEIGHBORHOODS ARE CHANGING TABLE 3-15 Regression Analysis of Change in Housing Units Without Central Heating 99 Sample Excluding Excluding Los Angeles Variable Los Angeles and New York Constant+.13 +.20 SMSA population change+.89 +.77 (4.2) (2~9) SMSA income change-.54 -.90 (3~3) (3~4) SMSA filtering-.86 -.78 (3~4) (2~6) Change in occupied housing stock +.18 (1.7) R2.37 ·49 NOTES: Figures in parentheses are t-ranos of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentages or percentage changes. Los Angeles neighborhoods omitted because of the climate. Unemployment and Labor Force Participation The analysis of unemployment and labor force participation indicates that these neighborhoods should be regarded as local sources of labor for a metropolitan market, rather than as isolated local labor markets. The ability and willingness of the neighborhood's residents to supply labor to work depend on their personal characteristics, such as age, skills, and preferences (whether a woman has a small child to care for, for instance). The willingness of employers to hire them depends on the demand for the firm's output in the aggregate, on the demand for the output of all firms in the metropolitan area. The analysis of neighborhood unemployment and labor force participation (Ibbles 3-16 and 3-17) is essentially a reduced form of this model; the SMSA unemployment rate is the proxy for the demand for labor, and labor force participation is measured by the fraction of the adult population that is not in the labor force. Both metropolitan demand and local supply contributed to the labor force changes. A rise in the SMSA unemployment rate raised the neigh- borhood unemployment rate, in all samples, on about a one-for-one basis. Several attributes of the neighborhood were related to the local unemploy- ment rate, particularly the growth in Hispanic population and changes in household composition, in different samples (Bible 3-16~. It is encouraging to note that the unemployment rate fell more in neighborhoods where the educational level of the population increased more, but this effect unfortunately disappeared when Los Angeles was excluded.

100 INNER-CITY POVERTY IN THE UNITED STATES TABLE 3-16 Regression Analysis of Changes in Neighborhood Unemployment Rate Sample Excluding A1179 Excluding Los Angeles Vanable Neighborhoods Los Angeles and New York Constant +.02 +.01 +.01 SMSA unemployment rate change +.75 +.90 +1.09 (6~7) (4.7) (4.7) Hispanic population change -.05 -.05 (3~3) (2~9) Change in median years of schooling -.53 (2.4) Mamed couple change -.05 (2.3) Pemale-headed household change +.03 (1~9) Change in occupied housing stock - +.06 (2.2) R2 .48 .35 .53 NOTES: Figures in parentheses are t-raiios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentage changes except the change in schooling, which is the change in the median number of years. Labor force participation was less affected by the SMSA unemployment rate and more by the neighborhood attributes: age and race consistently, schooling and household composition in some cases Gable 3-17~. In most of these neighborhoods, the unemployment rate rose over the decade and labor force participation declined. It is frequently argued that one important factor contributing to unemployment in urban ghettos is suburbanization; increasingly businesses and jobs are moving to the suburbs, leaving the urban poor, particularly minorities, unable to get to them in effect, unable to supply their labor. Solutions to the problem range from improving public transit systems to building subsidized housing in the suburbs. (For an exposition of this view, see Kasarda, 1986.) Data on job location within SMSAs, available in the decennial census, permit some investigation of this issue. Individuals report whether they worked the week that the census is taken the week of April 1 and where they worked: in the central business district, elsewhere in the central city, or in the suburbs. These responses were aggregated for the entire SMSA to obtain a picture of the changes in job location and included in the statistical analysis. Changes in job location for the central business district, the rest of the central city, and the central city as a whole, were included as separate independent variables in the analyses of unemployment and labor force participation. Generally, the findings indicate that job suburbanization has not mattered to the people in these poverty neighborhoods. None of the

HOW POVERTY NEIGHBORHOODS ARE CHANGING TABLE 3-17 Regression Analysis of Neighborhood Changes in Labor Force Participation (measured as percentage of adults not in the labor force) 101 Sample Variable _ _ . All 76 Neighborhoods Excluding Los Angeles Excluding Los Angeles and New York Constant Change in city's share of SMSA jobs SMSA unemployment rate change Elderly population change +.05 -.12 (3~5) +1.14 (4.7) +.53 (2~9) -.10 (1.6) Change in population of "other races" -.54 (2.6) Youth population change Hispanic population change Female-headed household change Change in median years of schooling R2 +.a7 -.10 (2.7) +1.00 (4.2) +.63 (3.4) -.12 (2.1) -.82 (3~9) +.01 (1.3) .55 -04 +.12 (1.4) +~03 (2.5) +.67 (2.7)  +.35 (1.6) -.17 (1.5) +.11 (2.5) .50 .52 NOTES: Figures in parentheses are t-ratios of the regression coefficients. All variables refer to the neighborhood unless "SMSA" is specified. All variables are expressed as percentage changes except Me change in schooling, which is the change in the median number of years. Lord Beach and Gary omitted. locational changes was significant in any of the unemployment regressions. Jobs have become more dispersed in most of these SMSAs, but that has not affected the unemployment rates in the poverty neighborhoods. The labor force regressions present a somewhat different picture. In the two larger samples, a shift in jobs from the city to the suburbs results in an increase in the proportion of neighborhood residents who choose to drop out of the labor force altogether. The effect is small, however. A decline of 17 percentage points in the city's share of SMSA jobs the largest for any city in this sample is associated with a 2 percentage point decline in the labor force participation rate. Moreover, the relationship disappears, and in fact is reversed, when New York and Los Angeles are excluded. The coefficient for the smallest sample is positive, and equally large in absolute value, but it is not conventionally significant. These results suggest that suburbanization of jobs is a problem for people living in poverty neighborhoods in New York and Los Angeles, but not in the other large eastern and midwestern cities. One might speculate that the problem is more serious in New York though the regression results do not particularly suggest it. Half the New York neighborhoods

102 INNER-CITY POVERTY IN THE UNITED STATES are in Brooklyn; the nearest suburbs are in Nassau County, with Queens in between. (There is only one poverty neighborhood in Queens.) Some Manhattan residents are close to New Jersey, but the census data for the borough as a whole show relatively little commuting in that direction. Most of the Los Angeles neighborhoods are close to the Civic Center and have good access to freeways. Several neighborhoods are just inside the city limits; Watts is essentially a peninsula surrounded by suburbs on three sides. This is about as far as one can go with the current data set. More detailed information on job location would be needed, as well as further tabulations of census tract data on automobile ownership. It would also be desirable to include other cities in the analysis. POLICY IMPLICATIONS: WHAT MIGHT HELP? Poverty neighborhoods clearly have persistent problems and they may be deteriorating. Public policy at best kept them from getting worse during the 1970s; it did not make them better. One purpose of this committee's study is to evaluate alternative policies that might help. In this concluding section, some past and present policies are considered in light of the changes and current conditions in these neighborhoods. It is convenient to categorize policies in terms of their orientation: Do they try to help people or try to help places? Some programs have tried to do both, but most have a primary focus on one or the other. Programs for the Poor Income transfers are increasingly important to the people in poverty neighborhoods, though they are not necessarily urban programs. Transfer payments probably account for a larger share of neighborhood income; certainly more residents have been receiving them. The data tabulated for this chapter do not distinguish between Social Security and welfare pro- grams, but given the changing age distribution and household composition in these neighborhoods, it is likely that both have become more important income sources. These transfers are surely helping the people, but they are probably not doing much to make the neighborhoods more desirable places to live. The Family Support Act of 1988 may help. This welfare reform legislation has several features intended to deter the formation of single- parent families. The law establishes mandatory child support by absentee fathers, including establishing paternity, automated tracking of support payments, and use of Social Security numbers. In addition, states can deny Aid to Families With Dependent Children to single mothers who are

HOW POVERTY NEIGHBORHOODS ARE CHANGING 103 minors, if they have moved out of their parents' home. These changes may reduce the proportion of female-headed families in these poverty neighborhoods. As a by-product, they would also probably reduce the demand for housing units; some married couples would exist in place of two separate households, and some young women would continue to live with their parents. The worst housing in these neighborhoods would drop out of the housing stock more rapidly. The Family Support Act also includes a `'workfare" component that consists of mandatory job training or schooling, as appropriate. Either could help some people in these neighborhoods to participate more effectively in the labor force. One of the few positive factors in these neighborhoods has been the improving educational level of the adult population, which apparently increased labor force participation and reduced neighborhood unemployment; this did not, however, translate into higher incomes in the neighborhoods. Unfortunately, the typical adult still had less than a high school education. Education and training will not completely reverse the decline in labor force participation, no matter how successful they are. Many residents are out of the labor force simply because they are old. A one percentage point growth in the elderly population translated into a one percentage point decline in labor force participation, and the growth in the elderly population in the average neighborhood accounted for about a quarter of the decline in participation. Single mothers accounted for another substantial share of the decline in participation, and they might benefit from welfare reform in the long run. Mothers with very young children would not have to participate in workfare, so the short-term effectiveness would depend on whether the full package of reforms successfully deterred single parenthood. Housing Housing programs are in an intermediate category; they attempt to help people and places at the same time. The decade of the 1970s witnessed unprecedented production of both private and subsidized housing. The subsidy programs may have helped improve the quality of life in the poverty neighborhoods. New low-income housing projects seem to have slowed the loss of population, and they may have given the residents more space to . . ve In. The housing programs of the 1970s had other purposes as well. At the beginning of the decade, Sections 235 and 236 had as one objective opening up the suburbs for the urban poor.4 They could have contributed 4 Most subsidized housing programs (including Section 235 and 236) are commonly known by the number of the section of the National Housing Act in which they are created, or else (as in

1(~4 INNER-CITY POVERTY IN THE UNITED STATES to the emptying of poverty neighborhoods. Apparently, however, they did not. The programs were large but short-lived, and suburban subsidized housing did not affect city housing conditions during the early 1970s (We- icher, 1982~. Section 8, in the last half of the decade, was even larger. The Section 8 New Construction program, at least during the 1970s, dispro- portionately served the white elderly poor, allowing them to live in decent neighborhoods. It might have contributed to the population decline in the poverty neighborhoods, but the program came so late in the decade that any substantial effect by 1980 is not likely. The Section 8 program evaluations to date have not addressed this issue. Whatever effects housing construction programs had during the 1970s, they will probably prove to have smaller effects during the 1980s. The Section 8 New Construction program was terminated in 1983 and no con- struction program has been passed to replace it, but even so, a substantial number of new units have been added to the stock during this decade. Moreover, the "pipeline" is long; like other federal low-income construc- tion programs, Section 8 New Construction projects often take several years to complete. Nonetheless, a larger share of the newly subsidized households in the 1980s live in existing housing and receive rental assistance under the Section 8 Existing Housing certificate program or the more recent voucher program. These programs serve many households who already live in de- cent housing but have a high rent burden, and they stimulate some modest improvements and perhaps some increased maintenance of the existing stock. The programs can also be used to move into better housing in a bet- ter neighborhood. The effect on the neighborhood depends basically on the general quality of housing to begin with. If the local private housing stock is generally decent, the programs help poor people afford the housing and promote neighborhood stability. If the stock is substandard, the programs help poor people move out of the neighborhood. The neighborhoods in this study probably fall more in the latter category. Several of these neighborhoods include traditional public housing projects. Public housing offers special problems and special opportuni- ties. Many urban projects consist of concentrations of the poorest and least skilled members of society. Some projects have been notoriously undermaintained, and many of the largest public housing authorities have been officially designated as "troubled" by the U.S. Department of Housing and Urban Development (HUD), meaning that they fail to manage their projects efficiently. At the same time, much of the public housing stock is , ~ . the case of Section 8) by their section number in the act adding them to the National Housing Act.

HOW POVERTY NEIGHBORHOODS ARE CHANGING 105 structurally adequate and decent; it is better than the private housing avail- able in many poverty neighborhoods. Moreover, public housing projects have more stable populations than poverty neighborhoods in general. Resident management and tenant ownership of public housing have been strongly advocated by the Bush administration and by HUD Secretary Jack Kemp. They may be one way to achieve neighborhood improvement. At first sight, this may seem an unlikely strategy; many public housing residents are very poor, and projects in many poverty neighborhoods consist of high-rise apartment buildings. Tenant ownership must perforce take the form of cooperative or condominium ownership, or ownership by a resident management corporation. Nonetheless, there have been some spectacular resident management success stories in urban public housing projects, and a number of projects have been converted to resident ownership since 1985, in a demonstration sponsored by HUD. Some badly deteriorated projects have been turned into attractive communities. The process of managing and owning a project has enabled the residents to acquire skills that they can use in other aspects of their lives. Since resident ownership and tenant management have become important components of housing policy, they have generated significant expressions of interest from public housing residents in many projects across the country. Tenant ownership typically requires financial support from the federal government. Some projects need to be rehabilitated. In addition, because the tenants are often very poor, they may need subsidies to meet the ongoing operating costs of the projects, at least for a time, just as they did while renting. Community Development Strategies It is hard to assess the potential of programs to help places rather than people in poverty neighborhoods. Since they are persistent slums, they are by definition places where local economic development either has not been tried or has not worked. If it had succeeded, the neighborhood would have moved out of poverty. There may have been some successful neighborhood development projects in low-income urban areas, but not many. A few census tracts in the cities under study moved out of poverty during the 1970s, but very few were in or near the poverty neighborhoods. The reason for the improvement cannot be ascertained from census data alone. The Washington, D.C., neighborhood mentioned earlier is an example of pure gentrification, rather than a public or private community development effort to help the poor. The major urban policy proposal of the Reagan and Bush adminis- trations has been the enterprise zone. The current zone proposal offers

106 INNER-C~ BERM IN THE UNFED STARS both capital and labor subsidies, in the form of tax incentives, to businesses located in areas with high unemployment. It is intended to promote en- trepreneurship and business formation, as well as job creation. The labor subsidy takes the form of a 5 percent refundable tax credit for the first $10,500 of wages paid to individuals working in the zone and having total wages of less than $20,000. The credit phases out between $20,000 and $25,000. A related policy is the subminimum training wage included in the minimum wage increase enacted in 1989. This may enable residents of poverty areas to find their first job and get a start on a better life, even if they are not immediately lifted out of poverty. At the same time, the overall minimum wage increase will tend to reduce employment of low- skilled poverty neighborhood residents, if anything. National Economic Policy and Poverty Neighborhoods The differences between the 1960s and 1970s and the connections between the neighborhood and metropolitan area economies illustrate the importance of national economic changes and macroeconomic policies. Real economic growth was substantial in the 1960s, and people in poverty neighborhoods fared better in real terms, though they lost ground relative to people in the rest of the SMSA Economic performance was weaker in the 1970s; the increase in family income for the nation as a whole was half as great. In most of the poverty neighborhoods in this study, real incomes declined. Macroeconomic policies had other neighborhood repercussions. Hous- ing production was much greater in the 1970s, in large part because of accelerating inflation. Early in the decade, a stimulative macroeconomic policy held interest rates down and encouraged record housing production. In the late 1970s, double-digit inflation fueled a speculative housing boom as part of a flight from financial assets to real ones. Geographically, the suburbs became more popular, the cities became less desirable, and the poverty neighborhoods emptied out. Inflation promoted filtering. Despite the strong record of economic growth during the l980s, it seems clear that the decade will be similar in key respects to the 1970s. The poverty rate for central cities was substantially higher in 1988 (the latest year available) than it was in 1979: 18.3 versus 15.7 percent. (Measured consistently, the rate was 16.5 percent in 1988 and about 14 percent in 1979.) While 1989 was a year of continued economic growth, the poverty rate surely did not decline to the 1979 level. The rate during the 1970s also increased; it was 13.4 percent in 1969. (Poverty neighborhoods are delineated after each decennial census on the basis of annual incomes reported for the preceding year.)

HOW POVERTY NEIGHBORHOODS ARE C~4NGING 107 The overall increase in central-city poverty in the 1980s is the net result of a sharp rise during the recession years from 1980 to 1982, and a steady but smaller decrease during each year of the economic recovery that continued through the rest of the decade. The official central-city poverty rate rose from 15.7 percent in 1979 the last full year of the cyclical recovery that began in 1975 to 19.9 percent in 1982. The rate rose substantially during the mild 1980 recession to 17.2 percent for the year. This is of some interest because the 1980 election marked a significant and controversial change in social welfare policy, and because public discussion of poverty trends during the decade often confuses the economic cycle that started at the beginning of 1980 with the political policy shift that started at the beginning of 1981. Although central-city poverty increased in both cycles, the increase during the Reagan administration (1980 1988) will probably be noticeably less than the increase during the decade of the 1980s (1979~1989~. The last two decades have seen a disturbing concentration of poverty in central cities. The overall poverty rate in 1988 was slightly lower than in 1969 (measured consistently), but the central-city rate was markedly higher. The logical inference from these data is that poverty neighborhoods in big cities expanded geographically during the 1980s, repeating the pattern of the 1970s. CONCLUSION: PROSPECTS AND POSSIBILITIES Looking at places rather than people is inherently discouraging. The hypothetical average person living in a poverty neighborhood was worse on in 1980 than 1970, in many ways. But any real individual probably did better than this hypothetical average. There were fewer people in the neighborhoods in 1980, and many of them may not have lived there in 1970. We tend to think in terms of neighborhoods because data exist for neighborhoods. Because "the neighborhoods" are worse off both relatively and absolutely, we think of their residents as an underclass. But this is potentially misleading, because of mobility. The slums have persisted, but many of the people have moved on, and we do not know whether they are better or worse off. As a society, we are really more interested in the well-being of people than places. In fact, the most important public policies toward povertr neighborhoods have not been place oriented. Explicit public policy has increasingly centered on income maintenance; implicitly, policy has en- couraged people to leave poverty areas. That is not a bad combination for the future. It should be abundantly clear by now that poverty neighbor- hoods are generally unattractive places to live. We should expect people to seek better neighborhoods, if they can. At the same time, we can

108 INNER-CITY POVERTY IN THE UNWED STATES probably improve current income maintenance programs so that they help beneficiaries find their own way out of poverty, and perhaps out of poverty neighborhoods. Part of the income maintenance and also part of the inducement to move could be a housing voucher program or an expansion of the Section 8 Existing Housing certificate program. Public housing, Section 8 New Construction, and other low-income construction programs have proven to be much more expensive than using the existing stock to house the poor. Public housing changes the neighborhood as well as the housing stock. It has slowed the rate of population decline in poverty neighborhoods and lowered the turnover. Improvements in neighborhood housing quality, however, have been small at best. Overall, the beneficial effects on the neighborhood, if there are any, are probably far too small to justify the expense. This conclusion is based on the objective neighborhood changes reported in the decennial census. This chapter has not tried to analyze the role of subsidized housing construction in creating an underclass or promoting a culture of poverty, which are more important but more nebulous issues. A declining but substantial minor fraction of the people in these neighborhoods are children (about one-third in 1980, down from about 40 percent in 1970~. The strategy of income maintenance for the current residents may not help the children get out of poverty when they grow up. They need more and better education. A campaign to discourage dropping out of school would help. Public schools in many poverty neighborhoods do not provide an adequate education, however. Improving the schools would make the neighborhoods more desirable places to live, but that is obviously a tall order. Education vouchers would let poor people send their children to better schools outside the neighborhood, which might keep relatively high-income households in the poverty neighborhoods when their children reach school age. That would promote neighborhood stability and perhaps neighborhood development. Whatever policies are followed, the prognosis is for further population decline. This need not be the fate of every poverty neighborhood; some can probably be turned around. But these neighborhoods seem to be attractive mainly to new immigrants in growing metropolitan areas. Some of the neighborhoods in the eastern and midwestern cities were ports of entry for immigrants from Europe until immigration was restricted after 1924. A few have drawn the new immigrants from Latin America; if the study were extended to include more southern and western cities, there would be more such neighborhoods. The Simpson-Mazzoli Act of 1988, if it is effective, will discourage new immigration in the future. Within a few years, the Los Angeles poverty neighborhoods are likely to start losing population as some of the new immigrants move up on the economic ladder and can afford better places to live. The poverty neighborhoods in the East, which

HOW POVERTY NEIGHBORHOODS ARE CHANGING 109 have seen some influx of Hispanics and Asians, will continue to emptier out, probably faster. ACKNOWLEDGMENTS The author benefited from discussions with Cicero Wilson and assis- tance from Carlton Henry and David Hover in the initial stages of the research, and from comments on an earlier draft by John L. Goodman, Jr., Daniel Weinberg, and the members of the Committee on National Urban Policy. Sole responsibility for any errors rests with the author. This study was supported by a grant from the Joyce Foundation to the American Enterprise Institute, and was completed while the author was in residence at the American Enterprise Institute. The opinions and conclusions are also those of the author, and not necessarily those of the American Enterprise Institute, the Joyce Foundation, or the U.S. Department of Housing and Urban Development. REFERENCES Bureau of the Census 1966 Characteristics of Families Residing in "Poverty Areas": March 1966 Cur rent Population Reports, Series P-23, No. 19. Washington, D.C.: U.S. Department of Commerce. 1973 Characteristics of Low-Income Areas. 1970 Census of Population, Vol. II, No. 9B. Washington, D.C.: U.S. Department of Commerce. Money Income and Poverty Status in the United States: 1988 (Advance Data from the March 1989 Current Population Survey). Series P 60, No. 166. Washington, D.C.: U.S. Department of Commerce. Butler, Edgar, et al. 1969 Moving Behavior and Residential Choice: A National Survey. National Cooperative Highway Research Program Report No. 81. Washington, D.C Greenberg, Michael R., and Thomas D. Boswell 1972 Neighborhood deterioration as a factor in intraurban migration: a case study in New York City. Professional Geographer 24(Februaty):11-16. Kasarda, John D. 1986 The regional and urban redistribution of people and jobs in the U.S. Paper prepared for the Committee on National Urban Policy, National Research Council, Washington, D.C. Lowly, Ira S., ed. 1983 Expe~imeniing wish Housing Allowances. Cambridge, Lass.: Oelgeschlager, Gunn & Hain. Nathan, Richard 1986 The Underclass Will It Always Be With Us? Paper prepared for a symposium at the New School for Social Research, November. National Housing Policy Review 1974 Housing in the Seventies. Washington, D.C: U.S. Department of Housing and Urban Development. 1989

110 Putnam, Israel 1973 INNER-CITY POVERTY IN THE UNITED STATES Poverty Neighborhoods in 105 Large Central Cities, A Special Tabulation From the 1970 Census. Photocopy. Available from John C. Weicher, U.S. Department of Housing and Urban Development. Weicher, John C. 1973 The Causes of Housing Improvement in Low-Income Neighborhoods Paper prepared for the National Housing Poligy Review. Available from the author. 1982 The Relationship Between Subsidized Housing Production and Loss Rates Wthin Metropolitan Areas. Contract Report No. 1484~1 to the U.S. De partment of Housing and Urban Development. Washington, D.C.: Urban Institute. 1987 Mismeasuring poverty and progress, CATO Joumal 7(Winter):715-730.