## Abstract

We highlight the paradoxical implications of decadal reclassification of U.S. counties (and America’s population) from nonmetropolitan to metropolitan status between 1960 and 2017. Using data from the U.S. Census Bureau, we show that the reclassification of U.S. counties has been a significant engine of metropolitan growth and nonmetropolitan decline. Over the study period, 753—or nearly 25% of all nonmetropolitan counties—were redefined by the Office of Management and Budget (OMB) as metropolitan, shifting nearly 70 million residents from nonmetropolitan to metropolitan America by 2017. All the growth since 1970 in the metropolitan share of the U.S. population came from reclassification rather than endogenous growth in existing metropolitan areas. Reclassification of nonmetropolitan counties also had implications for drawing appropriate inferences about rural poverty, population aging, education, and economic growth. The paradox is that these many nonmetropolitan “winners”—those experiencing population and economic growth—have, over successive decades, left behind many nonmetropolitan counties with limited prospects for growth. Our study provides cautionary lessons regarding the commonplace narrative of widespread rural decline and economic malaise but also highlights the interdependent demographic fates of metropolitan and nonmetropolitan counties.

## Introduction

Urbanization has continued apace in the United States (Gibson 2010). Indeed, growth of the nation’s metropolitan population—its absolute size and percentage share—has accelerated over the past century (Johnson and Lichter 2019).1 Today, the growing influence of metropolitan America is evident in national political debates and governance, cultural values and mass opinion, and regional and community investments and innovation (Lichter and Brown 2011). Some rural areas, of course, continue to grow and prosper, but others have been relegated to the demographic, economic, and political sidelines. America’s small towns and thinly settled rural areas, especially in Appalachia, the Great Plains, and the Delta, have faced protracted economic difficulties—poverty, diminished employment, job instability, and stagnant wage growth—as well as population decline over the past several decades. A new “geography of despair” is now expressed in increasingly concentrated rural poverty rates and declining life expectancy caused by rising midlife mortality from suicides and drug overdoses, especially from abuse of fentanyl and other opioids (James et al. 2019; Monnat and Brown 2017; Thiede et al. 2018). The 2016 presidential election and its aftermath have further placed the political spotlight on disadvantaged rural people and places.2

We show that the spatial and social boundaries that typically separate rural and urban America are highly fluid and ambiguous (Frey 2017; Lichter and Ziliak 2017). This fundamental point is illustrated here by (1) documenting how national and subnational geographic patterns of population growth and decline since 1960 have been influenced by nonmetropolitan-to-metropolitan reclassification; (2) describing the demographic processes that have produced this population change; and (3) highlighting how reclassification has affected the sociodemographic composition of nonmetropolitan areas over time, diminishing the stock of human capital and the potential for growth. Our analyses show how county reclassification—from nonmetropolitan to metropolitan status—has increased the proportion of nonmetropolitan counties experiencing chronic depopulation and has hastened the process of cumulative rural disadvantage, in part because many high-growth nonmetropolitan counties are redefined as metropolitan after each decennial enumeration. This demographic paradox poses many interpretative and data-analytic challenges that impede appropriate evaluations of rural social and economic change and possibly undermine appropriate public policy interventions (Murdock et al. 2012; Schaeffer et al. 2013).

## Background

### Reclassification and Growth and Decline Processes

In rural America, depopulation has emerged as a widespread demographic phenomenon over the past century (Johnson and Lichter 2019). The U.S. nonmetropolitan population peaked overall in 1940, when 75 million people or roughly 57% of all Americans lived in small towns, in the open countryside, or on farms (Gibson 2010). Only 10 years later, in 1950, a slight majority of all Americans lived in metropolitan areas. Recent population estimates from the U.S. Census Bureau (2018) indicate that only 46 million now live in nonmetropolitan areas. This figure represents about 14% of the total U.S. population—an all-time low. Since 1940, nonmetropolitan areas have “lost” nearly 30 million people. Much of the loss was due to reclassification of nonmetropolitan counties, which “transitioned” to metropolitan status.

The period since 2010, however, marks a significant new inflection point historically in America’s evolving settlement system. For the first time, nonmetropolitan counties experienced overall absolute population decline. Between 2010 and 2016, nonmetropolitan counties declined by more than 190,000 people, representing a 0.4% loss (Cromartie 2017).4 Prior to 2010, most nonmetropolitan areas, as defined at the beginning of each decade, increased in population size during the decade despite rural net out-migration to metropolitan areas. This occurred only because natural increase—the difference between births and deaths—exceeded losses from rural out-migration (Johnson and Lichter 2016). This is no longer the case. Chronic net out-migration of young adults has over time sapped the demographic vitality and reproductive potential of many nonmetropolitan areas, which is revealed today in unprecedented population aging and incipient population decline (Glasgow and Brown 2012; Johnson and Lichter 2008). Mortality rates have climbed upward, and historically low fertility has reduced prospects for growth (Johnson 2011; Johnson and Winkler 2015). For much of nonmetropolitan America, natural increase has given way demographically to natural decrease, which has reinforced the downward population spiral caused by continuing net out-migration from nonmetropolitan counties.

The lessons learned from past and contemporary patterns of metropolitan and nonmetropolitan population change are much less straightforward than they seem. That is, population change results from (1) endogenous population growth and decline processes within a constant universe of counties defined either as metropolitan or nonmetropolitan, and (2) metropolitan reclassification due to urbanization (i.e., the transfer of metropolitan and nonmetropolitan counties from one category to the other). By definition, metropolitan and nonmetropolitan growth and decline processes are highly interdependent. The OMB defines metropolitan areas based on counties or county equivalents (e.g., in New England) (OMB 2010). Metropolitan counties—all 1,167 of them—include so-called central counties with urbanized areas (principal cities and densely settled areas of 50,000 or more people) and outlying counties linked to core counties as indicated by observed labor force commuting patterns. Nonmetropolitan is a residual category, representing all counties that do not meet the metropolitan size and density criteria. After each decennial census, counties are subject to reclassification based on changes in population and commuting patterns during the preceding decade. Overall metropolitan and nonmetropolitan growth depends on population growth in the same universe of counties—endogenous growth—and on transfers of counties (usually but not always) from nonmetropolitan to metropolitan status.

The literature on reclassification is small and badly outdated (Brown 1979; Fuguitt et al. 1988; Nucci and Long 1995). For example, Fuguitt et al. (1988) disaggregated metropolitan decadal population gains over 1950–1980 into three components of growth: (1) internal growth within counties initially classified as metropolitan, (2) expansion by the addition of so-called outlying counties (i.e., adjacent nonmetropolitan counties) to existing metropolitan areas, and (3) emergence of entirely new metropolitan areas. As urbanization progressed, historical declines in the absolute size of the nonmetropolitan population resulted in large part from reclassification rather than endogenous nonmetropolitan decline. Fuguitt et al. (1988) showed that reclassification hastened the metropolitanization process, a result Johnson et al. (2005b) confirmed during the 1980s and 1990s. Recently, Goetz et al. (2018) showed that metropolitan areas grew much more than nonmetropolitan counties between 1969 and 2015, and that reclassification played an important role in this process. They claimed that the “story of rural malaise is just as much a story of some nonmetro counties being remarkably successful” (Goetz et al. 2018:100). Whether this claim is true requires a careful evaluation of changing patterns of growth and decline among nonmetropolitan counties that transition to metropolitan status over successive decades. What is known is that it was only after 2010 that nonmetropolitan counties overall experienced endogenous declines in population (Cromartie 2017). Much of this growth of metropolitan America occurred at the periphery of existing metropolitan areas (e.g., suburbanization) rather than from the formation of entirely new metropolitan areas.

One clear implication of previous studies is that many “left behind” nonmetropolitan counties may no longer be demographically self-sustaining, either from net in-migration or natural increase (Johnson and Lichter 2019; Thiede et al. 2017). Rural depopulation, observed for the first time overall in the post-2010 period, arguably reflects a demographic winnowing process that has unfolded over many decades. That is, many nonmetropolitan counties with the most potential for growth, perhaps because of their larger sizes or unique locational advantages (such as proximity to metropolitan employment centers), have been redefined over time as metropolitan. Other declining or slow-growth nonmetropolitan counties have been left behind in the urbanization process. This metropolitan selection process—one based on differential nonmetropolitan growth—has seemingly culminated over time in an increasingly disadvantaged universe of remaining nonmetropolitan counties (i.e., those with the least potential for future growth).

### The Paradox of Nonmetropolitan Growth and Composition

The conceptual boundaries that define nonmetropolitan and metropolitan counties are made increasingly ambiguous by reclassification from decade to decade. Recent reports of widespread nonmetropolitan decline and growing economic and social malaise are appropriate from a strict technical point of view (Lichter and Ziliak 2017). They accurately describe conditions of nonmetropolitan counties but only as currently defined (i.e., at the time of the last census). But such a narrow reading of the evidence arguably gives a rather incomplete or even misleading interpretation of evolving demographic conditions in rural America. Paradoxically, in the extreme, rural depopulation arguably could be viewed positively if more rural areas are urbanizing and, as a result, leaving the nonmetropolitan universe through population growth and economic prosperity. The current zero-sum nature of measuring metropolitan and nonmetropolitan growth and decline is inextricably linked to the reclassification of U.S. counties from one decade to the next (Frey 2017; Fuguitt et al. 1988). Rural people and places are brought into the metropolitan fold by reclassification, which itself is a result of nonmetropolitan net in-migration and natural increase (Johnson and Lichter 2019).

Metropolitan and nonmetropolitan counties represent two sides of the same coin. Highly selective reclassification not only affects the observed population growth rates of metropolitan and nonmetropolitan counties but also leads to possible changes in the rural population composition, which usually goes unrecognized (Brown 1979). For example, are slow declines or even increases in poverty rates in nonmetropolitan counties a consequence of reclassification of the most prosperous nonmetropolitan counties? Over time, some segments of the nonmetropolitan population (i.e., certain geographical areas or population subgroups, such as the highly educated or affluent) may be more likely to be reclassified as metropolitan. This means that nonmetropolitan populations—rich or poor, young or old, Black or White—become redefined as part of the metropolitan population without actually migrating. Indeed, they often reside in the same houses, neighborhoods, and communities. Reclassification reflects the process of urbanization, including the growth of the metropolitan principal cities and outward suburban expansion as well as increased commuting that results from economic restructuring. Nonmetropolitan counties in the path of urbanization are swept along in the process. As the boundaries separating nonmetropolitan from metropolitan are redrawn, metropolitan areas have encompassed more territory but also more people with social and economic characteristics that are different—more advantaged—from the people left behind in lagging nonmetropolitan counties. Reclassification thus has both quantitative and qualitative implications for understanding comparative growth in metropolitan and nonmetropolitan areas.

An early study by Brown (1979) illustrated this substantive point. More than 200 nonmetropolitan counties were redefined as metropolitan after the 1970 census, accounting for roughly 50% of all nonmetropolitan growth between 1960 and 1970. The transitioning nonmetropolitan counties grew by nearly 25% overall during the 1960s compared with only 4% among nonmetropolitan counties that were left behind. Brown (1979) found that as a result of reclassification, average education levels declined slightly in the remaining nonmetropolitan areas. Of course, compositional effects over a single decade are likely to be small but cumulate in much bigger compositional effects over several decades. This winnowing of human capital—through selective reclassification of people over consecutive censuses—may leave behind the most disadvantaged individuals in nonmetropolitan areas. From a demographic standpoint, selective reclassification from nonmetropolitan to metropolitan status has much the same consequences as chronic nonmetropolitan to metropolitan outmigration; it drains rural human capital. The challenges that many rural areas face today are rooted in America’s demographic past and in the selective reclassification of prosperous counties (and people) into the universe of metropolitan counties. To fully understand the demographic processes that underlay cumulative disadvantage requires analysis over a time horizon that spans several decades.

### Current Study

Our singular goal is to show that the process of growth and decline in metropolitan and nonmetropolitan areas is linked in fundamental ways to the metropolitan reclassification of nonmetropolitan counties over the 1960–2017 period. Our main hypothesis is that decade-to-decade quantitative (i.e., population size and growth) and qualitative (i.e., socioeconomic composition) changes have culminated over this 57-year period in substantial metropolitan growth at the expense of nonmetropolitan areas. We show that reclassification is driven by nonmetropolitan population change that results from both net migration and natural increase (i.e., the difference between births and deaths). In the absence of net in-migration and/or natural increase, nonmetropolitan counties are unlikely to grow sufficiently to meet the criteria necessary for reclassification to metropolitan status (see upcoming full discussion). This is an especially important issue today when nonmetropolitan population growth is at historical lows, exacerbated by widespread natural decrease and continuing outmigration to metropolitan areas (Johnson 2011; Johnson and Lichter 2019). Resuscitating population growth in lagging nonmetropolitan counties arguably has become more difficult than ever.

## Data, Methods, and Analytical Approach

### Data

County population counts and estimates provide the basic accounting units for our analyses of reclassification. They come from the decennial censuses for 1960–2010 and from the Census Bureau population and components of change estimates from 2010 to 2017 (U.S. Census Bureau 2018). Counties—all 3,146 of them—have several advantages that are useful for our purposes.5 Unlike cities and communities (i.e., places, as defined by the Census Bureau) or urban areas, most counties have historically stable boundaries suitable for tracking endogenous population growth over time. They also are the smallest unit for which the federal government reports comprehensive demographic data. Counties also are important local government units that serve as political and administrative actors to shape local growth and decline processes (e.g., economic development strategies, zoning, and taxation policy). We use the terms rural and nonmetropolitan interchangeably, as we do the terms urban and metropolitan.

Decade-to-decade population change is a function of a complex interplay between natural increase and net migration. County-level historical data on births, deaths, and net migration for 1970–2010 are available from the integrated age-specific net migration files developed by multiple research teams over the past 60 years (Winkler et al. 2013). We supplement these with Census Bureau estimates of population and demographic change for 2010 to 2017. County net migration is derived by the residual method; net migration is what is left when natural increase (births minus deaths) is subtracted from the total population change. These data provide insights into how the demographic components of population change have contributed to county reclassification from nonmetropolitan to metropolitan status.

Reclassification impacts not just the demographic characteristics of counties but also their human capital. We measure this by examining changes in per capita income, percentage of college graduates, percentage in poverty, and age structure. Per capita income data from 1970–2017 come from the Bureau of Economic Analysis (2019), education and poverty data for the same period are compiled from Census Bureau sources by the U.S. Department of Agriculture (USDA) (2019), and the age data are from the Census Bureau.

### Measuring Metropolitan Status

The number of counties identified as metropolitan is based on criteria delineated by the OMB. Following the 2010 decennial census, in 2013, there were 1,167 metropolitan counties and 1,979 nonmetropolitan counties. The basic concept of a metropolitan area (MA) has not changed significantly since its introduction in 1950. MAs consist of one or more central counties that contain an urban nucleus of 50,000 or more people, together with any outlying counties that are economically integrated with the core. Because of this consistency, especially the 50,000 urban place population thresholds, most of the changes in county status from nonmetropolitan to metropolitan, or vice versa, reflect on-the-ground patterns of urban growth and suburban expansion. This makes MAs a useful geographic unit—one that is inclusive of all U.S. territory and people and readily available for tracking the growth and expansion of metropolitan counties, often at the expense of nonmetropolitan counties.

Each decade, the MA criteria (or standards) are reviewed by OMB in consultation with federal statistical agencies (OMB 2010). Even though the basic concept of a MAs and the use of counties to delineate them have remained unchanged, the criteria used to define metropolitan areas have been periodically revised. Changing criteria can affect the metropolitan-nonmetropolitan status of counties, independent of underlying urbanization trends (Fuguitt et al. 1988). Changes in the criteria used to define metropolitan areas in the 1950s and 1960s were minimal but increased considerably thereafter (OMB 1998). Definitional changes in the 1980s and 2000s significantly reduced the number of nonmetropolitan counties. For example, the 1980s revisions eased the requirements to be designated as a central city, making it easier to form the core of a new metropolitan area. At the same time, stricter standards for measuring city-suburb economic integration made it more difficult for an outlying county to join an existing metropolitan area. These strict standards were greatly simplified following the 2000 census, which redefined many nonmetropolitan counties as metropolitan.

### Metropolitan Reclassification and Population Change

Reclassification is an integral aspect of the metropolitanization process. It reflects shifts in metropolitan status, with counties transferred either from nonmetropolitan to metropolitan status or, much more rarely, from metropolitan to nonmetropolitan status. Reclassification reflects demographic growth and decline processes as well as spatial expansion of urban areas; and as we described earlier, it is influenced by the criteria used by OMB to define metropolitan. Although the number of counties classified as metropolitan or nonmetropolitan has varied through the decades, the overall trend has not. After each census, the number of nonmetropolitan counties has declined, and the number of metropolitan counties has increased.

Because some metropolitan and nonmetropolitan counties are reclassified each decade, the usual approach is to track population using a fixed universe of metropolitan and nonmetropolitan counties over each decade (Fuguitt et al. 1988; Johnson and Lichter 2016). Metropolitan and nonmetropolitan counties are defined by OMB shortly after the start of each decade, when tabulations from the most recent census are released. To study metropolitan population change during the 1990s, for example, the population of metropolitan counties in 1990 and 2000 was based on the universe of metropolitan counties designated by OMB in 1993, after the results from the 1990 census were reported. Metropolitan population change—and its nonmetropolitan residual—were each based on the same universe of counties throughout the decade of the 1990s. For the 2000s, the universe of metropolitan and nonmetropolitan counties was revised in 2003 to take into account reclassification based on the results of the 2000 census, and the same metropolitan and nonmetropolitan counties followed over the decade. This conventional approach to studying decade-to-decade metropolitan and nonmetropolitan population change contrasts with a floating definition, which allows the universe of metropolitan and nonmetropolitan counties to change throughout the decade. In practice, this means that metropolitan population change—say, from 1990 to 2000—includes a different universe of metropolitan (nonmetropolitan) counties at the beginning and end of each decade. Population growth would reflect both endogenous growth (in the same counties) as well as growth and decline in counties that are reclassified.

To highlight the implications of reclassification, we divide counties into four groups using the 1963 metropolitan classification as a point of departure. The first group includes the 1,979 counties defined as nonmetropolitan in 1963 (based on the results of the 1960 census) that remained nonmetropolitan at the end of the study period (i.e., through the 2010 census and subsequent 2013 metropolitan classification). Counties that were defined as metropolitan in 1963 (based on the 1960 census) and remained metropolitan as defined in 2013 (based on the 2010 census) provide an important contrast to these nonmetropolitan counties. These 414 continuously metropolitan counties contain densely settled metropolitan core counties that include most of the nation’s older central cities as well as many older suburbs. They also include adjacent noncore counties that reflect city-to-suburb residential mobility of young families beginning after WWII.

Another group includes 389 counties that were classified as nonmetropolitan in 1963 but had been redefined as metropolitan by 1993 (based on the 1990 census). These early transition counties either were added to existing metropolitan areas or were part of entirely new metropolitan areas by 1993. They made the transition during the rural population turnaround of the 1970s or during the economic downturn of the 1980s. A second group of counties, defined as nonmetropolitan in 1993, transitioned to metropolitan status by 2013 (based on 2010 census results). These 364 late transition counties shifted from nonmetropolitan to metropolitan status during the rural rebound of the 1990s or the turbulent economic era of the 2000s, which included the economic boom of the early 2000s as well as the Great Recession and its immediate aftermath. These two groups of reclassified counties—early and late transition—represent the transfer of land and population from nonmetropolitan to metropolitan status. They are of interest because they reflect the impact of reclassification on population and demographic change in metropolitan and nonmetropolitan America.

### Analytical Approach

Our analyses proceed in three stages. First, we describe changes in the number of counties that have been redefined as metropolitan for each decade from 1960 to 2017. This allows us to gauge the implications of both endogenous change and reclassification on metropolitan and nonmetropolitan population growth since 1960. Second, we show that reclassification results from growth from both natural increase and net migration but that these components have shifted markedly over time in ways that work to the disadvantage of nonmetropolitan counties. Third, we examine the extent to which reclassification contributed to cumulative shifts (i.e., declines in human capital) in the social, economic, and demographic composition of the nonmetropolitan population.

## Results

### Reclassification and the Urbanization of Nonmetropolitan America

Our first objective focuses on the urbanization of rural areas to highlight the implications of reclassification for changing patterns of population growth in metropolitan and nonmetropolitan areas. This is an important goal. Between 1963 and 2013, the OMB reclassified 753 (24%) of all U.S. counties from nonmetropolitan to metropolitan status. Reclassification was geographically widespread but was more concentrated in the South and West, particularly among the late transitioning counties (Fig. 1). There was considerably less reclassification in the Great Plains. Reclassification was most pronounced on the periphery of existing metropolitan areas, although some entirely new metropolitan areas formed, often in regions of the country that originally had few metropolitan counties.

In the absence of reclassification, nonmetropolitan areas (based on a constant 1963 definition) would have gained considerable population since 1960. Indeed, the 753 counties shifting from nonmetropolitan to metropolitan status between 1963 and 2013 contained 69.5 million people in 2017 (Fig. 2). Among these, the 389 early transition counties contained considerably more residents (54 million) than the 364 late transition counties. The overall impact of reclassification was substantial. It shifted 24% of all U.S. counties and 21% of the nation’s population from nonmetropolitan to metropolitan status between 1963 and 2013. The Census Bureau’s recent population estimates show that the population in the remaining nonmetropolitan counties stood at only 46.0 million in July 2017, representing a new low of only 14% of the entire U.S. population. The clear paradox is that chronic declines in the absolute and relative size of America’s nonmetropolitan population have contributed to metropolitan growth (i.e., the urbanization of nonmetropolitan counties). In essence, a significant share of metropolitan growth is fueled each decade by reclassification.

The pace of reclassification has been uneven, with both the number of nonmetropolitan counties and residents reclassified to metropolitan status varying from decade to decade. Between 1963 and 1973, some 229 counties were reclassified from nonmetropolitan to metropolitan. Reclassification reduced the size of the nonmetropolitan population by 13.1 million people during 1960–1970, from 67.5 million using the 1963 metropolitan definition to 54.4 million using the 1974 metropolitan definition (Fig. 3). A net loss of counties and population occurred in each following period as well. Between 1973 and 1983, for example, a net of 88 counties were reclassified as metropolitan: these counties contained 8.6 million residents in 1983. An additional 101 counties were reclassified as metropolitan in 1993, which transferred 5.8 million residents from nonmetropolitan to metropolitan status. Another 7.3 million people, from 253 counties, became metropolitan between 1993 and 2003. These numbers declined between 2003 and 2013, when only 4.8 million rural people and 77 counties were reclassified from nonmetropolitan to metropolitan.6

The decadal changes evident in Fig. 3 clearly illustrate the immediate impact of reclassification, but they do not fully reflect the aggregate impact over time of reclassification. That is, nonmetropolitan America lost not only the residents transferred in the decade of reclassification (as discussed earlier) but also the cumulative population gains that these reclassified counties amassed in later years. These cumulative impacts were substantial, as illustrated on the right axis of Fig. 3, which compares the 2017 population of the counties that were reclassified between 1963 and 2013 with that of the counties that remained nonmetropolitan in 2013. For example, the 2,700 counties that were classified as nonmetropolitan in 1963 (based on the 1960 census) included 115.5 million people in 2017. In contrast, the counties still classified as nonmetropolitan today include only 46 million people, based on population estimates provided by the U.S. Census Bureau (2018). The cumulative loss of 69.5 million residents reflects the aggregate impact of metropolitan reclassification as of 2017.

The net effect of reclassification is striking. In 1970, 33.1% of the U.S. population resided in nonmetropolitan areas (nonmetropolitan + transition) with the remaining 66.9% in metropolitan counties (Fig. 4). By 2017, just 14% of the population was nonmetropolitan and 86% metropolitan (metropolitan + transition). Counties that were reclassified from nonmetropolitan to metropolitan accounted for all the proportional growth of the metropolitan population during that period. In fact, the share of the U.S. population residing in the original 1963 metropolitan counties diminished from 67% in 1970 to 65% in 2017. Counties that remained nonmetropolitan in 2013 contained 18% of the U.S. population in 1970, and this declined to 14% in 2017. Counties that transitioned from nonmetropolitan to metropolitan contained 15% of the U.S. population in 1970 but more than 21% in 2017. The substantive implications are striking: all the increase in the metropolitan share—from 67% in 1970 to 86% in 2017—was due to the reclassification of nonmetropolitan counties (i.e., they became metropolitan). The percentage of the U.S. population residing in counties that were already metropolitan in 1963 remained essentially unchanged during the period.

In sum, an important but usually unacknowledged reason for the substantial cumulative impact of reclassification is that nonmetropolitan counties reclassified as metropolitan grew faster than other groups of counties. In each decade between 1970 and 2017, the counties that transitioned to metropolitan early consistently grew at the highest rate, and the late transition counties usually had the second highest growth rate (Fig. 5). In contrast, except during the nonmetropolitan population turnaround of the 1970s, counties that remained nonmetropolitan in 2013 grew at the lowest rate. Counties defined as metropolitan throughout the study period generally grew more rapidly than counties that remained nonmetropolitan but less rapidly than the counties that transitioned from nonmetropolitan to metropolitan. Thus, the share of the population that was metropolitan grew not only because of endogenous population increase but also because fast-growing nonmetropolitan counties were being added to the population base of existing metropolitan areas or forming new metropolitan areas.

### Reclassification and the Components of Population Change

Our second objective is to illustrate how reclassification is the result of population gains from both net migration and natural increase. Population growth is greatest in those counties reclassified from nonmetropolitan to metropolitan, but how have the demographic components of population change—net migration and natural increase—affected reclassification and the metropolitanization process?

As shown in Fig. 6, net migration produced most of the population gains in both groups of reclassified counties. The importance of migration in these counties may not be surprising given that many were located on the periphery of existing metropolitan areas and benefited from suburbanization. However, there is a striking difference between the relative contribution of net migration and natural increase in these transition counties compared with those that were not reclassified. Natural increase produced more than two-thirds of overall population gain in both consistently metropolitan and consistently nonmetropolitan counties.

The metropolitanization of America is primarily driven by migration. In counties reclassified from nonmetropolitan to metropolitan, net in-migration fueled population gains, even during decades of turbulent demographic change. For example, during the rural turnaround of the 1970s and the rural rebound of the 1990s, net migration gains in counties that had transitioned, or soon would, from nonmetropolitan to metropolitan status far exceeded that in the counties that remained consistently metropolitan or nonmetropolitan (Fig. 7). Even during the 1980s, 2000s, and 2010–2017 periods, when nonmetropolitan gains were far more constrained, migration gains in the reclassified counties exceeded those in the consistently metropolitan or nonmetropolitan counties. It is important to recognize that a substantial share of this net migration gain in transition counties came from existing metropolitan counties, particularly older urban cores and suburbs, as well as from immigration.

Gains from natural increase were much smaller in continuously nonmetropolitan areas than in nonmetropolitan counties that were reclassified as metropolitan or in counties defined as metropolitan throughout the study period. In part, the changing patterns of natural increase are a function of historical net migration trends. With some exceptions, nonmetropolitan counties have consistently lost young adults of reproductive age to metropolitan areas (Johnson et al. 2005b, 2015). Over time, the sustained outflow of young adults from most nonmetropolitan areas has diminished the population of childbearing age, while the older population has aged in place. The net impact has been fewer births and eventually more deaths in aging rural counties, resulting in less natural increase or—in an increasing number of nonmetropolitan counties—outright natural decrease.

### Reclassification and Shifts in the Stock of Human Capital

What are the implications of reclassification for human capital in metropolitan and nonmetropolitan areas? Our third objective addresses the question of whether reclassification has also affected population composition in ways that have exacerbated metropolitan-nonmetropolitan differences in the accumulation of human capital between 1970 and 2017. Table 1 tracks the trajectories of change for four key indicators of human capital: age, education, per capita income (in constant 2012 dollars),7 and poverty. These data illustrate our main point: continuously nonmetropolitan counties are socioeconomically disadvantaged in comparison with counties that were reclassified as metropolitan.

Our analyses of other indicators of human capital reinforce this main conclusion. Specifically, counties defined as nonmetropolitan throughout the study period (i.e., today’s nonmetropolitan counties) had lower per capita incomes than counties that transitioned to metropolitan status or that were always defined as metropolitan. In 2017, inflation-adjusted per capita income (reported in 2012 dollars) was $37,432 in counties that remained nonmetropolitan. In contrast, per capital income was$44,445 in early transition counties, $39,216 in late transition counties, and$53,272 in continuously metropolitan counties. The early transition counties already had 16.6% higher income per capita in 1970 than did the counties that remained nonmetropolitan. The initial income advantage was more modest among nonmetropolitan counties that transitioned later. Clearly, it was counties that already had higher incomes—the economic “winners”—that were more likely to transition from nonmetropolitan to metropolitan status, leaving behind nonmetropolitan counties with less income.

In light of the results for income, it is not surprising that the percentage in poverty was higher in 2017 for today’s nonmetropolitan counties (15.9%) than for counties that transitioned early (12.3%), transitioned later (14.3%), or were classified as metropolitan throughout the study period (12.8%).9 These differences in poverty were not simply the result of an economic boost associated with transitioning to metropolitan status (and thus becoming more integrated with the nation’s metropolitan economy). In 1970, 21.3% of residents in continuously nonmetropolitan counties were in poverty. By comparison, poverty rates were 44% lower in early transition counties (15.3%), 17.5% lower in late transition counties (19.6%), and 79% lower in continuously metropolitan counties (11.0%). The chronic poverty observed in today’s nonmetropolitan counties is clearly in part due to a selective winnowing process since 1970.

Finally, reclassification also left behind an older population in nonmetropolitan America: 19% of the current nonmetropolitan resident population is 65 years or older, compared with 16.4% in the early transition counties and 18.3% in the late transition counties. The continuously nonmetropolitan counties were already older in 1970 than those that would transition to metropolitan status over the ensuing decades. As we have argued here, population aging often depletes demographic vitality, diminishes the stock of human capital, and reduces the likelihood of future growth.

The winnowing of human capital resulting from reclassification is evident when current and reclassified counties are combined to measure the effect of reclassification on current nonmetropolitan demographic composition. The education loss was substantial. Among counties that were initially nonmetropolitan (including those that transitioned to metropolitan), 24% of residents over age 25 had a college degree in 2017, compared with 19.3% among those that remained nonmetropolitan in 2013. Income losses were significant as well. Per capita income diminished from $40,937 to$37,432. And, the nonmetropolitan poverty rate increased by 2 percentage points, from 14% to nearly 16%. The net transfer of income and education reflected in the transition of counties from nonmetropolitan to metropolitan has been substantial over the past half century. Reclassification transferred 19% of the college graduates and 19% of earned income from nonmetropolitan to metropolitan America between 1970 and 2017. This presents significant challenges for today’s remaining nonmetropolitan counties as they compete to attract and retain population and economic activity.

## Discussion and Conclusion

The decadal reclassification of U.S. counties (and America’s population) from nonmetropolitan to metropolitan status historically has been a significant indicator of urbanization in America (Cromartie 2017; Fuguitt et al. 1988; Johnson et al. 2005a). Here, we show that many fast-growing nonmetropolitan counties reached a population size and functional level required for reclassification to metropolitan status each decade over the past 50 years. Ongoing metropolitanization of nonmetropolitan America is a demographic paradox—one not often fully appreciated in the social science literature. Many of the fastest growing nonmetropolitan counties are being redefined each decade as metropolitan. Other nonmetropolitan counties, especially those in remote and isolated rural areas, have been left behind by an increasingly urban settlement system marked by the emergence and growth of new metropolitan areas and the peripheral expansion of existing metropolitan areas. Ours is thus a tale of two nonmetropolitan types: those counties that have successfully transitioned to metropolitan status and those that have been left behind. To be sure, some of the remaining nonmetropolitan counties continue to grow and prosper, but many face significant demographic and economic challenges in the future.

Our study provides important lessons for future research on the U.S. settlement system. One is that reclassification has been an important dimension of metropolitan growth and nonmetropolitan decline over the past half century. This is not always appreciated. Commentators often contrast the substantial numeric and proportional population gains of metropolitan areas to the far more modest gains and diminishing share of the population that is nonmetropolitan. Yet, our analyses, first and foremost, show that the 753 nonmetropolitan counties that were redefined as metropolitan between 1963 and 2013 shifted nearly 70 million residents from nonmetropolitan to metropolitan America by 2017. This represents 21% of the entire 2017 U.S. population. In fact, all the proportionate growth in metropolitan America from 67% of the population in 1970 to 86% in 2017 resulted from metropolitan reclassification. Growth in these reclassified counties has been substantial, fueled by sustained net migration as well as by significant natural increase, which can be viewed as a second order effect of in-migration of young adults of reproductive age.

A second important lesson is that nonmetropolitan population decline should be viewed, at least in part, as a story of success. Declines since 1970 in the size of the nonmetropolitan population—from 67 million to 46 million in 2017—resulted paradoxically from rapid population growth in nonmetropolitan counties leading them to be subsequently reclassified as metropolitan. For decades, the remaining nonmetropolitan counties experienced modest population growth because they had enough natural increase to offset ongoing net out-migration. This historical pattern may now have come to an end. Between 2010 and 2017, nonmetropolitan counties, as currently defined, experienced absolute population decline for the first time in America’s history (Cromartie 2017; Johnson 2019). Although some nonmetropolitan counties—especially those proximate to metropolitan areas or those attracting amenity or lifestyle migrants—continue to gain population, the number of such counties at risk for reclassification is diminishing by the decade. This means that reclassification is less likely in the future to fuel the metropolitanization process. It also means that many of the remaining nonmetropolitan counties today face difficult challenges to renewed population growth through new in-migration and fertility.

The impacts of reclassification are not limited to population change alone. Indeed, many nonmetropolitan counties with the most potential for growth and economic development have been siphoned off over time into existing or newly defined metropolitan areas. This suggests a third important lesson: nonmetropolitan areas that have transitioned to metropolitan status typically include higher percentages of college-educated people, higher per capita incomes, lower poverty rates, and lower percentages of older people. Reclassification represents a highly selective process of urbanization. Over time, urbanization—at least the way it is typically measured and reported—exacerbates the deficit of human capital in nonmetropolitan America and limits the prospect of future economic growth. This confirms our central argument.

Our results also provide some cautionary methodological lessons about how best to track population change and accurately characterize urbanization and rural decline. Should metropolitan boundaries be held constant over time or float from decade to decade as a result of reclassification? The former uses a constant universe of metropolitan and nonmetropolitan counties over the study period, and the latter updates metropolitan and nonmetropolitan county classifications throughout the study period. For example, in the March Current Population Survey, explaining trends in nonmetropolitan employment, education, or poverty is potentially affected by reclassification. We show that both nonmetropolitan demographic trends and trends in education, income, and poverty are highly sensitive to the effects of reclassification, but that this fact has not always been appreciated in previous research. At a minimum, our findings are a call for scholars and policymakers to consider the impact of reclassification in their comparisons of metropolitan and nonmetropolitan America. Our work also underscores the need for greater public access to both fixed and floating definitions of metropolitan and nonmetropolitan people and places.

In conclusion, our study emphasizes the difficulties in appropriately defining nonmetropolitan/metropolitan areas and tracking social and demographic trends (Lichter and Ziliak 2017; Wunderlich 2016). On the one hand, our results seemingly temper some of the current pessimism about the state of rural America, especially if successful nonmetropolitan areas are no longer counted as nonmetropolitan because of population growth and successful economic development (Fallows 2019; Van Dam 2019). We are much less sanguine about the many thinly populated and declining areas left behind in rural America, where the prospects of future population and development seem more limited than ever.

## Acknowledgments

The authors gratefully acknowledge John Cromartie of the Economic Research Service of the USDA for his contribution to the early analytical work on this project. Kenneth Johnson’s research was supported by an Andrew Carnegie Fellowship from the Carnegie Corporation of New York and by the New Hampshire Agricultural Experiment Station in support of Hatch Multi-State Regional Project W-4001 through joint funding of the National Institute of Food and Agriculture, U.S. Department of Agriculture, under award number 1013434, and the state of New Hampshire. Barbara Cook of the Carsey School of Public Policy provided GIS support. The content is solely the responsibility of the authors and does not necessarily represent the official views of the agencies supporting their research.

## Authors’ Contributions

Both authors contributed equally to the conception, design, and execution of the project, including drafting and editing the final manuscript. Johnson was responsible for the acquisition and management of data and carried out the analyses in consultation with Lichter. Both authors read and approved the final manuscript.

## Data Availability

All the data used are publicly available from the sources listed in the Data section of this article.

## Compliance With Ethical Standards

### Ethics and Consent

No ethical approval or consent was required for this study because the data are all available from public sources.

### Conflict of Interest

The authors report no conflict of interest.

## Notes

1

We use the terms rural and nonmetropolitan as well as metropolitan and urban interchangeably here.

2

A substantial majority of nonmetropolitan residents voted for Donald Trump over Hillary Clinton, seemingly registering their dissatisfaction with the political status quo and urban-centric political concerns and public policy (Scala and Johnson 2017; Wuthnow 2018).

3

A 2015 National Academy of Sciences report on the “Workshop on Rationalizing Rural Area Classifications” highlighted this central point (Wunderlich 2016). Throughout U.S. demographic history, rural people and places have become part of the urban population through rural-to-urban migration but also through reclassification by OMB as nonmetropolitan populations are redefined as metropolitan.

4

Recent data suggest that nonmetropolitan areas have resumed growth in the past two years, although they have lost population over the entire period from 2010 to 2018 (Johnson 2019).

5

The number of counties will vary slightly from analysis to analysis because of boundary changes that complicate longitudinal analysis. Over the past century, a few new counties have been added, others have had boundary changes, and Virginia has introduced the concept of independent cities.

6

There are minor differences between the decadal reclassifications reported here and the early and late transition counties reported earlier. A modest number of decadal reclassifications were temporary. For example, a few counties reclassified as metropolitan in 1973 reverted to nonmetropolitan status in 1983. Such counties would be included in the decadal changes but not in the early and late transition classification, which delineates change over longer periods.

7

There is considerable disagreement about the most appropriate income measure to define economic well-being (Katz 2012). Katz noted that alternative income measures differ in terms of the sources of income (earnings or non-earned income, such as pension income) and their sensitivity to demographic factors, such as household size or number of working adults. Here per capita income is used. The correlation between the per capita income and median family income, for example, is quite high (.8), so overall patterns evident in one measure will be reflected in the other.

8

Detailed analysis of the longitudinal patterns of education and the other three human capital indicators (not included here) reveals that the significant gap between the continuously nonmetropolitan and the transition counties already existed in 1970 and remained relatively stable between 1970 and 2017. For example, the percentage of college graduates was 44% higher in the early transition counties than in continuously nonmetropolitan counties in 1970. In 2017, the gap was 48%. For per capita income, the percentage income difference between the early transition and continuously nonmetropolitan was 16.6% in 1970 and 21% in 2017. Thus, it was not becoming metropolitan that created the substantial differences in 2017 but rather the counties that transitioned had higher levels of human capital to begin with.

9

To maintain consistency with data provided by the Economic Research Service of the USDA (2019), we calculate the poverty rate as the number of people below the poverty line divided by the total population. This is not consistent with the U.S. Census Bureau calculation of the poverty rate, which is the number of people below the poverty line divided by the population for whom poverty status is known. As a result, our measure is lower than the official poverty rate. For example, in 2017, the official poverty rate was 14.6% compared with our reported percentage in poverty of 13.2%.

## References

References
Brown, D. L. (
1979
).
Metropolitan reclassification: Some effects on the characteristics of the population in metropolitan and nonmetropolitan counties
.
Rural Sociology
,
44
,
791
801
.
Bureau of Economic Analysis
. (
2019
).
Personal income by county, metro and other areas
Cromartie, J. B. (
2017
).
Rural areas show overall population decline and shifting regional patterns of population change
(Amber Waves report).
Washington, DC
:
U.S. Department of Agriculture, Economic Research Service
Fallows, J. (
2019
,
May
28
).
The rural-urban divide is more complicated than you think
.
The Atlantic Magazine
Frey, W. H. (
2017
).
The fading of city-suburb and metro-nonmetro distinctions in the United States
. In T. Champion, & G. Hugo (Eds.),
New forms of urbanization: Beyond the rural-urban dichotomy
(pp.
67
88
).
New York, NY
:
Routledge
. 10.4324/9781315248073-4
Fuguitt, G. V., Heaton, T. B., & Lichter, D. T. (
1988
).
Monitoring the metropolitanization process
.
Demography
,
25
,
115
128
. 10.2307/2061481
Gibson, C. (
2010
).
American demographic history chartbook: 1790 to 2010
Glasgow, N., & Brown, D. L. (
2012
).
Rural ageing in the United States: Trends and contexts
.
Journal of Rural Studies
,
28
,
422
431
. 10.1016/j.jrurstud.2012.01.002
Goetz, S. J., Partridge, M. D., & Stephens, H. M. (
2018
).
The economic status of rural America in the President Trump era and beyond
.
Applied Economic Perspectives and Policy
,
40
,
97
118
. 10.1093/aepp/ppx061
James, W., Cossman, J., & Wolf, J. (
2019
,
January
21
).
The ultimate inequality: Vast mortality disparities across U.S. regions
.
N-IUSSP on-line news magazine
Johnson, K. M. (
2011
).
The continuing incidence of natural decrease in American counties
.
Rural Sociology
,
76
,
74
100
. 10.1111/j.1549-0831.2010.00036.x
Johnson, K. M. (
2019
).
Rural America growing again due to migration gains
(Carsey Data Snapshot).
Durham
:
University of New Hampshire, Carsey School of Public Policy
. 10.34051/p/2020.353
Johnson, K. M., Field, L. M., & Poston, D. L. (
2015
).
More deaths than births: Subnational natural decrease in Europe and the United States
.
Population and Development Review
,
41
,
651
680
.
Johnson, K. M., & Lichter, D. T. (
2008
).
Natural increase: A new source of population growth in emerging Hispanic destinations in the United States
.
Population and Development Review
,
34
,
327
346
. 10.1111/j.1728-4457.2008.00222.x
Johnson, K. M., & Lichter, D. T. (
2016
).
Diverging demography: Hispanic and non-Hispanic contributions to U.S. population redistribution and diversity
.
Population Research and Policy Review
,
35
,
705
725
. 10.1007/s11113-016-9403-3
Johnson, K. M., & Lichter, D. T. (
2019
).
Rural depopulation: Growth and decline processes over the past century
.
Rural Sociology
,
84
,
3
27
. 10.1111/ruso.12266
Johnson, K. M., Nucci, A., & Long, L. (
2005
).
Population trends in metropolitan and nonmetropolitan America: Selective deconcentration and the rural rebound
.
Population Research and Policy Review
,
24
,
527
542
. 10.1007/s11113-005-4479-1
Johnson, K. M., Voss, P. R., Hammer, R. B., Fuguitt, G. V., & McNiven, S. (
2005
).
Temporal and spatial variation in age-specific net migration in the United States
.
Demography
,
42
,
791
812
. 10.1353/dem.2005.0033
Johnson, K. M., & Winkler, R. (
2015
).
Migration signatures across the decades: Net migration by age in U.S. counties, 1950–2010
.
Demographic Research
,
22
,
1065
1080
. https://doi.org/10.4054/DemRes.2015.32.38
Katz, A. J. (
2012
).
Explaining long-term differences in census and BEA measures of household income
(BEA Working Paper).
Washington, DC
:
Bureau of Economic Analysis
Lichter, D. T., & Brown, D. L. (
2011
).
Rural America in an urban society: Changing spatial and social boundaries
.
Annual Review of Sociology
,
37
,
565
592
. 10.1146/annurev-soc-081309-150208
Lichter, D. T., & Ziliak, J. P. (
2017
).
The rural-urban interface: New patterns of spatial interdependence and inequality in America
.
Annals of the American Academy of Political and Social Science
,
672
,
6
25
. 10.1177/0002716217714180
Monnat, S. M., & Brown, D. L. (
2017
).
More than a rural revolt: Landscapes of despair and the 2016 presidential election
.
Journal of Rural Studies
,
55
,
227
236
. 10.1016/j.jrurstud.2017.08.010
Murdock, S. H., Cline, M., & Zey, M. (
2012
).
Challenges in the analysis of rural populations in the United States
. In L. J. Kulcsar & K. J. Curtis (Eds.),
International handbook of rural demography
(pp.
7
15
).
Dordrecht, the Netherlands
:
Springer
.
Nucci, A., & Long, L. (
1995
).
Spatial and demographic dynamics of metropolitan and nonmetropolitan territory in the United States
.
International Journal of Population Geography
,
1
,
165
181
. 10.1002/ijpg.6060010205
Office of Management and Budget (OMB)
. (
1998
).
Alternative approaches to defining metropolitan and nonmetropolitan areas
.
Federal Register
,
63
,
70526
70561
.
Office of Management and Budget (OMB)
. (
2010
).
2010 standards for delineating metropolitan and micropolitan statistical areas
.
Federal Register
,
74
,
37245
37252
.
Scala, D. J., & Johnson, K. M. (
2017
).
Political polarization along the rural-urban continuum? The geography of the presidential vote, 2000–2016
.
Annals of the American Academy of Political and Social Science
,
672
,
162
184
. 10.1177/0002716217712696
Schaeffer, P. V., Kahsai, M. S., & Jackson, R. W. (
2013
).
Beyond the rural–urban dichotomy: Essay in honor of Professor A. M. Isserman
.
International Regional Science Review
,
36
,
81
96
. 10.1177/0160017612449981
Thiede, B., Brown, D. L., Sanders, S. R., Glasgow, N., & Kulcsar, L. J. (
2017
).
A demographic deficit? Local population aging and access to services in rural America, 1990–2010
.
Rural Sociology
,
82
,
44
74
. 10.1111/ruso.12117
Thiede, B., Kim, H., & Valasik, M. (
2018
).
The spatial concentration of America’s rural poor population: A post-recession update
.
Rural Sociology
,
83
,
109
144
. 10.1111/ruso.12166
U.S. Census Bureau
. (
2018
).
Annual estimates of the resident population: April 1 2010 to July 1 2017
U.S. Department of Agriculture (USDA)
. (
2019
).
Data sources for poverty and education
[County-level data sets].
Washington, DC
:
U.S. Department of Agriculture, Economic Research Service
Van Dam, A. (
2019
,
May
24
).
The real (surprisingly comforting) reason rural America is doomed to decline
.
The Washington Post
Winkler, R. L., Johnson, K. M., Cheng, C., Voss, P. R., & Curtis, K. J. (
2013
).
County-specific net migration by five-year age groups, Hispanic origin, race and sex 2000–2010
(CDE Working Paper 2013-04).
:
University of Wisconsin–Madison, Center for Demography and Ecology
.
Wunderlich, G. S. (
2016
).
Rationalizing rural area classifications for the economic research service: A workshop summary
.
Washington, DC
:
.
Wuthnow, R. (
2018
).
The left behind: Decline and rage in rural America
.
Princeton, NJ
:
Princeton University Press
. 10.2307/j.ctvc7787j

## Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.