Paper presented at the annual meeting of the Population Association of America, in Cincinnati, April 3, 1993
The views expressed are attributable to the authors and do not necessarily reflect the views of the U.S. Bureau of the Census
The attached paper was delivered at the annual meeting of the Population Association of America, in Cincinnati, April 3, 1993. The paper's text and tables generally refer to metropolitan areas (MAs) as defined by the Office of Management and Budget (OMB) effective December 31, 1992. These MAs, which were defined on the basis of 1990 MA standards and 1990 census data, subsequently received an additional updating by the OMB (also based on 1990 standards and data) effective June 30, 1993.
The 1993 updating resulted in creation or restoration of a number of primary metropolitan statistical areas within consolidated metropolitan statistical areas, restoration of one metropolitan statistical area, restoration or addition of several counties to existing MAs, restoration of a number of central cities, and changes in some MA titles.
Current (1993) MA definitions and central cities are presented in the OMB's Bulletin 93-17, which is available on paper or disk from the National Technical Information Service (NTIS) at (703)487-4650. The NTIS accession numbers for the bulletin are: PB 93-192-664 (paper version); PB 93-505-816 (word processor file, 5 1/4" disk); and PB 93-505-824 (word processor file, 3 1/2" disk). A short summary of the changes effective in 1993 that can be used with Bulletin 93-17 is available from our office ((301)763-5158). Also, an alphabetical listing of the 1993 MAs that provides 1991, 1990, and 1980 population data is available as CPH-L-145 from the Bureau of the Census at (301)763-5002. Finally, extensive 1990 data for the 1993 areas are presented in the 1990 census supplementary report Metropolitan Areas as Defined by the Office of Management and Budget, June 30, 1993 (CPH-S-1-1), which is available from the U.S. Government Printing Office under the stock number 003-024-08738-3.
We have not attempted to reflect the 1993 MA definition updating in population figures in the attached paper. On the next page, however, there is a revised version of the paper's Table 1, which provides a historical MA inventory.
Table 1. Number and Components of Metropolitan Areas: 1952 to 1992
For the fourth time since their establishment for the 1950 census, the official geographic definitions of metropolitan areas (MAs) underwent their decennial updating at the end of 1992. With the updating, the number of MAs in the U.S. fell, because the number of newly defined areas was less than the number of mergers of existing areas. The percentage of the nation's total 1990 population in MAs rose about 2 points, from 77.5 to 79.4, through the redefinition. There also was an increase in the percentage of the nation's metropolitan population accounted for by the largest MAs.
This paper summarizes the 1992 changes in metropolitan geography and their relationship to the population growth of MAs since the previous, post-1980 updating. It examines the extent to which the changes in definitions may reflect significant new trends in U.S. metropolitan structure, and concludes by outlining plans for a comprehensive review of the concepts underlying metropolitan and related statistical areas.
Although they have received widespread use by the private sector and in scholarly work, MAs are established and maintained by the Office of Management and Budget (OMB) to provide a standard geographic classification for use by federal statistical agencies. MAs include metropolitan statistical areas (MSAs), consolidated metropolitan statistical areas (CMSAs), and primary metropolitan statistical areas (PMSAs). The Census Bureau provides geographic expertise and technical support to OMB for defining MAs.
The general concept underlying MA definitions is that of a core area containing a large population nucleus together with adjacent communities having a high degree of economic and social integration with that core. This concept views the metropolitan area as a functional entity that is larger than the continuously built-up or urbanized area around the principal city or cities.
As operationalized in the MA statistical standards, each MSA generally comprises the core county or counties containing the principal city and its adjacent urbanized area, together with additional counties that have significant commuting-to-work ties with the core counties.1 The commuting level required for inclusion of these additional counties is variable: a flow to or from the core counties that is equal to 15 percent of a county's resident workers total qualifies that county as outlying if it is relatively densely populated and urban, for example, but a higher commuting level is required for a less dense and/or more rural county (holding other characteristics constant).
In those cases where an area meets the requirements to be an MSA and also has a population of one million or more, the MA standards detail procedures for determining whether major component areas can be identified within the entire area. If recognized, these component areas are PMSAs, and the entire area becomes a CMSA. For many statistical purposes it is convenient to treat the metropolitan universe as comprising the MSAs and CMSAs.
Other sections of the standards prescribe which places qualify as central cities, provide guidelines for deciding whether neighboring urban centers should be treated as one MA or separate ones, and specify area titles.
In limited specific situations, OMB seeks local opinion before arriving at a final definition or title for an MA. One such situation is the assignment of an outlying county that has approximately equal (and qualifying) commuting ties to different MAs. Another example is the identification of PMSAs, once they have met the required statistical criteria. Also, determination of PMSA and CMSA titles as well as titles of MSAs can involve local opinion in certain circumstances.
Although the general concept of MAs reflected in the official definitions has remained much as in 1950, the standards by which the MAs are defined have been revised several times. Thus, the updating of definitions after each decennial census has reflected modifications in the standards as well as actual changes in the extent of metropolitan development. The most extensive revisions to the standards took place in 1980 for use in the 1983 updating of definitions based on the 1980 census.
The current MA standards appeared in the Federal Register on March 30, 1990, after a period of review and public comment. Changes from the previous (1980) version of the standards were minor. The term "metropolitan area" was adopted as the collective term, replacing MSA, which is still the proper term for most individual metropolitan areas. An adjustment in the rules for qualifying counties containing a portion of an urbanized area resulted in adding six counties that otherwise would not have qualified. Another change resulted in qualifying six additional central cities. Other changes in the standards involved such aspects as the titling of areas and generally did not affect area boundaries or central cities.
Revisions to the set of MAs have occurred in practically every year since 1950, most often to recognize an MA for a growing small city. However, the great majority of revisions in the metropolitan definitions have occurred on the basis of decennial census data. These census-based revisions have taken place two or three years after the census, too late to be reflected in most publications of that census. The time lag stems from the fact that commuting data from the census are a key ingredient for determining which counties qualify for inclusion. The question on place of work is part of the sample census questionnaire and is not processed immediately for that reason; in addition, the place-of-work responses must be coded to permit tabulation in terms of standard geographic areas.
After 1960, 1970, and 1980, OMB did recognize a number of new individual MAs soon after the census based on total population counts, and these new areas were reflected in the tabulations and publications of that census. Updating of existing areas then took place separately, in the third year after each of these censuses. In some cases, however, the new commuting data caused areas created in one year to be changed or even merged with others two years later. For this reason, OMB decided not to recognize any new areas in 1991 but to defer all updating until the 1990 commuting data could be reviewed. OMB's announcement of MA definitions based on 1990 data and standards was effective December 31, 1992.
Fewer MSAs and CMSAs.2 The total number of current MSAs and CMSAs in the U.S. is 268, 16 fewer than the number in effect for the 1990 census (Table 1). (The 275 MSAs and CMSAs of 1983 had increased by nine new areas to 284 by 1990.) The 1992 announcement expanded 16 CMSAs and MSAs by adding to them one or more formerly separate MSAs; 24 MSAs were folded into other areas in this way. The most notable of these combinations brought together a pair in many respects dissimilar, Washington and Baltimore, forming a metropolitan complex that ranks fourth in the nation in population. Boston expanded by adding no less than five previously separate MSAs.
Besides the mergers, another 56 areas added territory, while 13 areas lost territory, and seven both added and lost. Finally, nine new MSAs were established in 1992, and one area, Jackson, TN, was disqualified because it had been established based on a 1984 postcensal population estimate that was not confirmed by the 1990 census. The 18 new MSAs announced between 1984 and 1992 are in contrast to the more than 40 new areas established between 1973 and 1983, some of which reflected a broadening of the requirements for qualification.
There has been some trend on the part of the largest areas against becoming CMSAs with component PMSAs. The motivation seems to be the negative effects (perceived or genuine) for the core of the area to have a smaller population total as a PMSA than it would have if the whole area remained undivided as an MSA. The count of CMSAs reached its peak in 1983, when 22 of the 35 metropolitan areas of at least 1 million population had PMSAs. Four additional areas at that time could have had PMSAs identified but turned the option down. Two areas that did opt for PMSAs in 1983 quickly changed their minds and secured a return to MSA status the next year. As of 1992 there were 40 areas of more than 1 million 1990 population, of which 27 could have had PMSAs. There was no significant local support expressed for the option in nine cases, reducing the count of CMSAs to 18.
In the 1992 updating, 167 of the 284 areas of 1990, or 58.8 percent, had no change in boundaries, although some of these did have changes in titles or central cities. This is about the same share as in 1983, when 174 or 60.9 percent of the 286 areas of 1981 had no change in boundaries.
More MA components. There were 12 counties added to metropolitan territory from 1984 to 1990, nearly all as part of newly recognized MSAs. There were no changes in definitions in 1991. The 1992 additions to metropolitan territory consisted of 90 counties added to existing MAs (usually on the basis of strengthened commuting ties) and nine in newly established areas, plus 62 additions of cities and towns to the New England areas; 18 counties and three New England units dropped out. Although these changes increased the number of metropolitan component areas by more than 10 percent, the effect on metropolitan population was much smaller. Using 1990 census populations, the 1992 additions added 5.5 million population while the deletions amounted to 0.7 million, for a net increase of 4.7 million or less than 3 percent of the 1990 total metropolitan population.
Of the 90 counties added in 1992, 16 had been metropolitan previously but were dropped at an earlier date, most of them in 1983 when the rules for qualifying outlying counties were tightened somewhat. Generally, since their earlier deletion from MAs such counties have achieved higher population densities and/or higher commuting to cores sufficient to meet the specified thresholds for inclusion.
Geographic distribution of the inventory changes. Counties added to metropolitan territory since the last updating are scattered across much of the country rather than concentrated in any one region (Figure 1). The two largest concentrations of added counties reflect the expansion of the Washington-Baltimore area by eight counties and one independent city, and the Cincinnati area by five counties--one in Indiana, one in Ohio, and three in Kentucky.
That the five counties added to the Cincinnati area accounted for a smaller population than the single county added to Cleveland is a reminder of the variable geographic contexts of MAs. Besides Cincinnati and Cleveland, counties were added to several other Manufacturing Belt MAs, including Fort Wayne, Chicago, Detroit, Rockford, and St. Louis, even though none of these areas grew greatly in population during the decade; the Detroit and Cleveland areas, in fact, gained counties while losing population within constant boundaries.
The census division with the largest number of additions was the South Atlantic. Seven of the nine new MSAs announced in 1992 are in this division, with North Carolina and South Carolina accounting for five of the areas. In total, the South Atlantic division saw 28 counties added to existing MAs (including those added to the Washington-Baltimore area) and nine included in newly recognized MAs after 1983 (Table 2). However, four counties also dropped out of MAs in this division in 1992. The high counts of changes in this division in part reflect that its counties tend to be small in area. The Atlanta MSA, which already had 18 counties, added three more but lost one to bring its total to 20, more components than any other area outside New England except New York and Washington-Baltimore.
Additions to metropolitan territory in the West occur mainly when new MAs qualify. In this decade, however, a few geographically very large counties in the West have qualified for inclusion as outlying MA counties.
In a number of cases throughout the nation the additional counties have filled gaps in existing chains or groups of MAs. Additions in New York and Ohio established or widened connections between Syracuse, Rochester, Buffalo, Erie, and Cleveland. Metropolitan counties are now continuous from the Greenville-Spartanburg-Anderson MSA in western South Carolina to the Greenville MSA in eastern North Carolina. With the exception of two counties, MAs occupy the entire Florida coast from Jacksonville south and then north on the Gulf Coast past Tampa. Establishment of a new MSA comprising San Luis Obispo County, hitherto the largest nonmetropolitan county in the country, completed continuous coverage of the California coast from the Mexican border to well north of San Francisco. However, as persons familiar with any of these areas will recognize, the fact that county-defined MAs are continuous is by no means the same as saying that urbanization or even suburbanization is continuous in these areas.
Of the 4.7 million net increase in 1990 MA population accounted for by the areas added in the 1992 announcement, 0.7 million was in the Northeast, 1.0 million in the Midwest, 0.8 million in the West, and 2.2 million in the South.
The primary determinant of the geographic extent of an individual MA is the territory from which it draws a significant share of workers; thus, changes in commuting patterns may bring significant changes in metropolitan boundaries. The number of component counties of MAs has risen much more rapidly since 1950 than the number of separate MAs, as many individual MAs have expanded from an original one-county definition to include two, three or more counties (Table 1).
The share of all workers who commute to a different county, which was 12 percent in 1954 and about 21 percent in 1980, had risen to 24 percent by 1990. Review of the 35 metropolitan areas with more than 1 million population in 1980 shows their intercounty commuting share at an even higher 27.5 percent, compared with 24.8 percent in 1980, but intercounty commuting rose more rapidly outside these major areas, from 16.9 percent to 20.2 percent of all workers.
The chief central cities of these 35 MAs continued as a group to gain jobs in the 1980s, but at the same time more than 80 percent of additional jobs in these areas were outside the central cities. Still, the central cities remain a dominant destination for commuters, with 34 percent more jobs than they have resident workers. In spite of major gains in jobs in suburbs, the suburban portions of every one of the 35 MAs still had a net outflow of commuters in 1990, indicating that suburban job growth has not yet caught up with rapid present or past suburban population growth.
Net commuting into the 35 major areas increased from less than 300,000 in 1980 to nearly 800,000 in 1990, using the boundaries of these areas as determined on the basis of the 1980 census. This was an indication that additional territory had developed significant commuting links in the intervening decade, and indeed much such territory has now been added to these areas by the updated MA definitions.
Examination of growth patterns for the largest MAs affords a perspective on the relative importance of different sources of metropolitan population growth while at the same time focuses on a part of the MA size spectrum that has accounted for an increasing share of the metropolitan population. With the 1992 redefinition, there were 40 metropolitan areas (MSAs and CMSAs) with at least 1 million population (1990) each. These MAs, which we will call "major metropolitan areas" (MMAs) in this discussion, accounted for 53.4 percent of the nation's population and two-thirds of the metropolitan population. (When 1991 population estimates were released earlier this year, one additional MSA had reached 1 million.)
Population growth. Table 3 shows the size and growth of MMAs on two bases. The left-hand block gives the totals for areas over 1 million as reported in each census. The right-hand block presents totals based on metropolitan boundaries determined by applying the current MA standards (those adopted in 1990 and applied in 1992) to data from the earlier censuses.3 The total MMA population is higher in the redefined series than in the census series, mainly because the redefined series includes counties added to the MMAs on the basis of that census; such counties generally had not been added officially in time to be included in the census series.
Regardless of the definitional approach used, the MMA category has undergone vast expansion in the last 50 years, during which the U.S. arguably has become a nation of metropolis dwellers. Applying the current standards, in 1940 the eleven MMAs of more than 1 million would have had 33.8 million population, just one quarter of the U.S. total. By 1990, the count of MMAs had increased to 40, and their total population had nearly quadrupled, to 132.7 million. The gross population gain of the MMA category since 1940 thus totals nearly 100 million, or about 85 percent of the nation's total population growth in that period.
Components of MMA growth. Individual MMAs grow through natural increase (with very rare exceptions) and also experience positive or negative population change through net migration. In addition, MMAs typically add population by expanding geographically over time. For any individual MMA, the extent and pace of geographic expansion is portrayed only in rough terms by definitions in terms of whole counties. An MMA may add two or three counties in one decade, and none in the next as suburban development occupies more and more of the recently added territory without spilling over into any additional counties. Still, for the entire set of MMAs such irregularities generally smooth out and provide a reasonably accurate picture of the extent to which the set of MMAs expands geographically.
For the set of MMAs as a whole, there is a further geographic component of change, the net additions of areas newly qualifying through reaching the threshold population of 1 million. It should be noted that the geographic components can work both ways; counties can drop out of MMAs (as two did in 1992), and, rarely, an area that had reached the 1 million threshold can decline to below that size.
Table 4 specifies the extent of each of these components of change for the MMAs during the five decades since 1940. The table is based on the series of definitions according to the current standards, shown in the right-hand block of Table 3. The "residual (including net migration)" category for a given geographic area represents the portion of that area's total intercensal change not accounted for by natural increase. Thus, besides net migration into or out of the area, it includes any net differential in coverage of the later census versus the earlier one.
The territorial expansion element is defined conservatively, comprising only the population of the added territory at the start of the decade it was added and not including any natural increase or net migration for the added territory during that decade. In other words, this treatment assumes that all of the territory added to MMAs, and any new MMAs, were qualified and added an instant after the start of the decade.
The major sources of MMA change have shown a broadly consistent breakdown over the 1940-90 period, with the 1970s a notable exception. For example, except in the 1970s natural increase within the starting MMA boundaries has consistently accounted for 29 to 35 percent of MMA growth; for the 1980s this component accounted for 34.3 percent. The residual (including net migration) category for initial boundaries fell from 21.5 percent in the 1940s to 12.9 percent in the 1960s, and then dropped sharply to only 2.3 percent in the 1970s before rebounding; for the 1980s, residual net migration contributed about 16 percent of all MMA growth.
Of the 25 million overall increase in MMA population during the 1980s, territorial growth accounted for 43 percent even on the conservative basis adopted. This share was around the average of the two preceding decades. Interestingly, the territorial component saw only a moderate decline across the exceptional decade of the 1970s; in other words, the population of territorial additions fell sharply in proportion as total MMA population change fell.
Within the territorial change category, however, the contribution from mergers of smaller MAs has shown a substantial increase over time, from the 5 to 6 percent range of 1940-60 to more than 15 percent in the 1980s. Two developments help account for this increase. First, the expansion of MMA commutersheds has meant that these areas more frequently impinge on neighboring, smaller MAs; and, second, the increase in the number of MAs of all sizes has increased the likelihood that an expanding MMA will have neighbors of metropolitan size to impinge upon.
The category of new MMAs--those areas newly past 1 million--saw its contribution to growth increase from about 23 percent in the 1940s to nearly 30 percent in the 1960s, when the number of MMAs grew from 22 to 32. In the 1970s and 1980s the contribution from new MMAs has returned to about the level of the 1940s. As it turns out, the lowered share from the qualification of new MMAs has about been balanced by the increased propensity of existing MMAs to absorb smaller areas near by.
The announcement of the new MA definitions for the 1990s brought one major set of activities towards closure and at the same time signaled new work on another--the consideration of alternative approaches to identifying and representing U.S. metropolitan/nonmetropolitan settlement patterns. The focus in this work is on the future, with operationalization of a new approach coming in time for use with the 2000 census data.
On the initiative of OMB, the Census Bureau began work in late 1989 on the Metropolitan Concepts and Statistics Project (MCSP). The project's mission is to reexamine the basic concepts underlying metropolitan and related statistical areas and propose alternative approaches for federal government use. During the project's first year, Census Bureau staff and two working groups of individuals from outside the federal government identified a set of ten questions that became the centerpiece of MCSP's research agenda. In addition to the overall concern with conceptual underpinnings, questions for study raised the issues of the fundamental geographical units or building blocks that would be used in identifying the entities of the settlement system under a proposed new approach, and the criteria by which the geographic units would be aggregated to create standard statistical areas. Other questions concerned the kinds and quality of data available for delineating alternative statistical areas, the frequency with which the areas would be updated, the possible role for local opinion, and how alternative approaches would generate data to satisfy different uses and users. Advice from the working groups also focused attention on the great desirability of an approach that would include the entire territory of the nation, not just its large centers.
The Census Bureau established joint statistical agreements (JSAs) with four universities in 1991. The principal researchers for these JSAs were: John S. Adams (University of Minnesota); Brian J.L. Berry (University of Texas at Dallas); Richard L. Morrill (University of Washington), and a team of William H. Frey (University of Michigan) and Alden Speare, Jr. (Brown University). The JSAs required that the researchers all address the same set of ten questions stemming from the discussions of the working groups in order to produce comprehensive, parallel approaches that could be compared in straightforward fashion.
Tee Census Bureau will publish the research results as a report in Current Population Reports, Series P23, next year. The report will show consistent concerns that a new approach (1) improve representation of those settlement system changes that have taken place over the last several decades, and (2) be attentive to the differences between morphological and functional measures in defining statistical areas.
Beyond these points, the approach of having each of the research efforts consider the same set of questions produced commonalities but also a rich range of choices. Looking at the primary geographic units to be employed, for example, finds that two papers argued for continued reliance on counties, one argued for use of county subdivisions (and for preparation of a county-based alternative), and the fourth argued for use of five-digit ZIP Codes. Assembly of these units into the proposed kinds of statistical areas was to rely on journey-to-work data in some cases but also on patterns of media penetration and structure density in other cases. One paper used population density criteria in constructing its principal statistical areas. A like range of views obtained in response to other questions as well.
Much work remains to be done before any new approach could become practicable. The five researchers provided us with promising directions with their excellent work conducted on shoestring budgets, but the assignment was one of exploration, without much application. We anticipate that there will be many public forums for discussion of these proposals and other work that stems from them later in the decade.
Much interest has been expressed in this project from many quarters during the past couple of years. We will be anxious to hear your comments.
Metropolitan areas are an important and widely recognized component of the nation's statistical infrastructure. Since MA standards were first applied to 1950 census data, they have received regular review and updating. The MAs and their increasing share of the total population have portrayed the nation's changing settlement patterns while providing a standard set of areas for reporting a wide array of data produced by the federal statistical system. The 1992 announcement of new MA definitions represents the fifth and latest installment in this effort. Attention now turns to whether the evolving settlement system, continued demands for high-quality geographically referenced data, and increased technological capacities indicate the superiority of some alternative approach. Proper treatment of this issue will consume much effort before the 2000 census.
1 In the New England States, MSA, CMSA, and PMSA definitions are in terms of subcounty minor civil divisions (cities and towns) and are based on analogous rules. Alternative, county-based New England MAs (called New England county metropolitan areas--NECMAs) are available for use with county-level data.
2 Puerto Rico MAs are not included in the following counts and discussion. Effective with the 1992 MA announcement there are three MSAs, one CMSA, and three PMSAs in Puerto Rico.
3 The data in this table for 1940 through 1980 were developed for a paper presented at the Association of American Geographers' annual meeting of 1988. The data have not yet been reviewed to determine whether the modest changes in the MA standards adopted in 1990 would have affected any of the MMA definitions for 1940 through 1980. We hope to complete a historical redefinition for MAs of all sizes by applying the current standards to earlier census data. For New England, the historical comparison uses county versions of the MMAs instead of the official city-and-town definitions because the latter are not yet available for earlier dates following the current standards.
Figure 1. Metropolitan Area Definition Changes: 1984-92 (GIF 118k)
Table 1. Number and Components of Metropolitan Areas: 1952 to 1993 (6k)
Table 2. Distribution, by Division, of Counties Added To and Deleted From Metropolitan Territory: 1984-92 (3k)
Table 3. Major Metropolitan Areas as Officially Defined and as Defined Under Current Standards: 1940-1990 (5k)
Table 4. Population Change by Source, Major Metropolitan Areas: 1940-1990 (10k)