Occasionally U.S. Housing Market Conditions uses this space to familiarize its readers with new data on housing or urban affairs. Past issues have discussed the property owners and managers survey, the Low Income Housing Tax Credit database, and the public use samples for Fannie Mae and Freddie Mac. This issue describes the state of the cities data systems (SOCDS), a new and convenient way to obtain basic economic and demographic data on cities and their suburbs.¹

Content and Format

SOCDS pulls together data from several sources into one easy-to-use format; provides information at the central cities, suburbs, and total metropolitan area levels of geography; and adjusts metropolitan level data to match the most recent definitions of metropolitan boundaries.²

SOCDS contains information from the 1970, 1980, and 1990 decennial censuses on population counts and population characteristics such as race, income, poverty status, education, employment status by place of residence, employment by industry and occupation, housing counts, and housing by tenure. The population counts are updated to 1996 with estimates from the U.S. Census Bureau. The SOCDS contains census data for all 539 metropolitan central cities and their associated metropolitan areas. For New England States, users can choose whether metropolitan areas follow the standard MSA/PMSA definition or the New England County Metropolitan Area (NECMA) definition. Data for the "suburbs" are calculated as the data for the metropolitan area less the sum of the data for all central cities (if any) in the metropolitan area.

From the Bureau of Labor Statistics (BLS), SOCDS contains counts of the labor force, employed persons, and the unemployment rate by place of residence for user-selected month and years between 1990 and 1998. BLS uses statistical techniques to produce these estimates based on data from the current population survey. SOCDS contains data for all U.S. central cities and metropolitan areas for which BLS publishes the data. "Suburbs" is computed as a residual, which means that for those metropolitan areas for which BLS does not publish data on one or more central cities, the definition of suburbs will differ from the definition used in SOCDS census data tables. In these cases, "suburbs" will include the omitted central cities.

SOCDS also contains special tabulations by the Census Bureau of the Standard Statistical Estab-lishment List, the source of data for the County Business Patterns publications.³ County Business Patterns Special Extracts (CBPSE) contain data for 114 central cities and their associated 101 metropolitan areas for 1993, 1994, and 1995.⁴ These include the 100 largest central cities, plus 14 additional cities so each State is represented. In New England States, metropolitan areas follow the NECMA definition. The "suburbs" in each metropolitan area are defined as the metropolitan area total less the sum of data for all the central cities for which data are available.

SOCDS data are organized first by source (decennial census, BLS, or CBPSE), then by State or metropolitan area, by city, and by topic. Thus a reader interested in population counts for Macon, Georgia, would first click on "historical data from the 1970, 1980, and 1990 Census," select Macon from a list of metropolitan areas or Georgia from a list of States, then Macon from a list of cities, and finally select "total population" from a selection of data topics.

The last section of this essay explains how to access SOCDS over the Internet. For researchers interested in comparative analysis across a large number of central cities or metropolitan areas, the data sets HUD used to construct SOCDS will be available at this Internet site in the near future.

Illustration

This section selects Pittsburgh, Pennsylvania, and Portland, Oregon, from the metropolitan areas profiled in the "Regional Activity" section of this issue of U.S. Housing Market Conditions to illustrate the data available in SOCDS.

The simplest and most accurate gauge of a city's or a metropolitan area's long-term economic performance is its population. The general pattern during the past 40 years has been for metropolitan areas to experience positive growth and to grow faster than their central cities. Many, but by no means most, central cities actually lost population during this period.

Pittsburgh's population fell by 32.6 percent from 1970 to 1996, while Portland's population grew by 25.9 percent.⁵ In both cases, the average annual change was greater (less negative in Pittsburgh's case) during the 1990 to 1996 period. The Pittsburgh metropolitan area also lost population during this period, while the Portland-Vancouver metropolitan area grew by 63.1 percent (table 1).

For this particular comparison, the long-run connection between population changes and economic fortunes can be seen by looking at the employment data. At the metropolitan level, particularly for isolated metropolitan areas such as Pittsburgh and Portland, there is likely to be a close association between job growth and employment growth, despite the fact that one is defined by place of work (jobs) and one by place of residence (employment). The same is not true at the city level because commuting can obscure where new jobs are being created; for example, jobs can increase in the city but employment increases in the suburbs.

Table 1. Population in 1970, 1990, and 1996

Table 2. Employment, 1970 and 1990

Between 1970 and 1990, employment grew only 8.3 percent in the Pittsburgh metropolitan area while growing 77.8 percent in the Portland-Vancouver metropolitan area. An interesting aspect of this difference in employment growth relates to manufacturing. In the Pittsburgh metropolitan area, manufacturing employment declined by 50 percent from 1970 to 1990, while in the Portland metropolitan area, it grew by 57.9 percent (table 2).

CBPSE data provide a more recent picture of job growth for the two areas. Here U.S. Housing Market Conditions takes advantage of data not currently on SOCDS. HUD has 1991 jobs data on 77 central cities and their metropolitan areas. When the Census Bureau finishes calculating the SSEL numbers for the remaining 37 cities, HUD will add the 1991 data to SOCDS.

Through the first half of the 1990s, Pittsburgh lost more than 6,660 jobs while its metropolitan area added more than 28,000 jobs. During the same period, Portland added more than 47,000 jobs and its metropolitan area added close to 100,000 jobs. In both central cities and their associated metropolitan areas, job performance was markedly stronger during the recovery than during the recession (table 3).

SSEL data also provide two other useful measures of recent economic activity: the number of establishments and annual payrolls. Both series show similar results over the 1991 to 1995 period: modest declines in the city of Pittsburgh, some growth in the Pittsburgh metropolitan area, and solid growth in both the city of Portland and the Portland-Vancouver metropolitan area.

To this point, SOCDS paints sharply contrasting pictures of the economic fortunes of Pittsburgh and Portland during the past three decades. At the city level, the long-run trends are completely different, with Pittsburgh declining and Portland prospering. Even at the metropolitan area level, Pittsburgh has seen only modest job growth from 1970 to the mid-1990s, and population has decreased.

Table 3. Change in the Number of Jobs, 1991-93 and 1993-95

Table 4. Unemployment Rates, August 1998

SOCDS can show how these patterns have affected residents. The most recent unemployment rates (August 1998) show that population and labor force have adjusted to these trends.⁶ The Pittsburgh and Portland metropolitan areas have almost identical unemployment rates, and Pittsburgh actually has a lower unemployment rate than Portland does (table 4).

Data on median incomes and poverty rates show the differences in economic trends for these two areas. In 1997 dollars, median incomes are higher in both 1969 and 1989 in Portland and its metropolitan area than in Pittsburgh and its metropolitan area (table 5).

Between 1969 and 1989, real median income rose 5.1 percent in the Portland-Vancouver metropolitan area compared with only 1.7 percent in the Pittsburgh metropolitan area. Although real incomes declined in both cities, the rate of decline was faster in Pittsburgh (7.6 percent compared with 2.1 percent in Portland).

Although both areas followed the national pattern of higher poverty rates in 1989 than in 1969, the increases were greater in Pittsburgh and its metropolitan area than in Portland and its metropolitan area (table 6).

Table 5. Median Income, 1969 and 1989 (in 1997 dollars)

Table 6. Poverty Rates, 1969 and 1989 (in percent)

The decennial census components of SOCDS provide important insights into the changing demographic compositions of these areas. SOCDS provides race and ethnicity data only from the 1980 and 1990 censuses because the racial and ethnic characterizations in the 1970 census do not match those in the 1980 and 1990 censuses.

The city of Pittsburgh has a substantially higher minority population than Portland does, and the minority share of population grew in both cities between 1980 and 1990, with the minority share growing faster in Portland. By 1990, the minority population in Pittsburgh city was predominately black, with only a small share of other race (non-Hispanic) and Hispanics. In Pittsburgh, the number of blacks in the central city actually declined between 1980 and 1990 by approximately 5 percent, but the number of whites declined much more (approximately 16 percent). While non-Hispanic blacks are the largest minority in the city of Portland, other race (non-Hispanic) is a close second and Hispanics are third. At the metropolitan level, the minority share has grown faster in the Portland area. Although the Portland-Vancouver metropolitan area had a smaller minority share in 1980, the reverse was true by 1990 (table 7).

The statistics on immigrants are particularly interesting. While the foreign-born share of the population in both Pittsburgh and its metropolitan area has been declining, the opposite is true for Portland and its metropolitan area. The explanation for this difference probably includes the better drawing power of Portland's superior economic performance during recent decades and the substantially greater growth of the population of other race (non- Hispanic) in Portland (table 8).

Table 7. Percentage Minority, 1980 and 1990

Table 8. Percentage of Foreign Born, 1970, 1980, and 1990

From 1970 to 1990, single-parent families grew rapidly as a share of all families with children throughout the Nation. Pittsburgh and Portland were no exceptions to this trend. The city of Pittsburgh had a higher percentage of single-parent families in 1990 than the city of Portland did, but the reverse was true for their respective metropolitan areas (table 9).

SOCDS also contains some basic housing information, such as tenure and vacancy status. The cities of Pittsburgh and Portland had homeownership rates above 50 percent in 1990. The trend in vacancy rates is more interesting. The loss of population in the city of Pittsburgh is clearly visible in its high and growing housing vacancy rate (table 10).

Table 9. Percentage of Single-Parent Families, 1970, 1980, and 1990

Table 10. Housing Vacancy Rates, 1970, 1980, and 1990 (in percent)

SOCDS Versus Spotlights

As the illustration shows, SOCDS documents well the economic and demographic trends that shaped Pittsburgh and Portland coming into the 1990s. More recent economic data trace employment and job trends through the middle to late 1990s. These data set the context in which current economic events, such as those related in the "Spotlight on" sections in this issue, unfold.

But data alone can do only so much to describe city conditions. For many conditions of interest, reasonably current data either do not exist or are available only with a substantial lag. For example, jobs data from the Census Bureau are available now only through 1995. Local sources are needed to fill this vacuum and the spotlights attempt to do this. In particular, the spotlights focus on local housing market conditions.

In addition to more timely and detailed information, the observations made by HUD's field economists in the spotlight sections give a sense of the local mood. The Pittsburgh spotlight clearly attests to the concern at the State and local levels about the area's long-run trend. The spotlight also reveals a sense of optimism that is certainly lacking from the longer term data.

How To Find SOCDS

Users can locate SOCDS by going to the HUD USER site on the Internet (http://www.huduser.gov) and clicking "Data Available from HUD USER." (SOCDS is not the same as the "State of the Nation's Cities Database," another useful database found at this site.) The Web address for SOCDS is http://socds.huduser.gov/.