Demographic Distress in the Heartland: Exploring Patterns of Growth and Decline

Katherine J. Curtis White, University of Washington
Jerald R. Herting, University of Washington

**Please note that Jerald R. Herting, University of Washington, is co-author on this paper. The online submission form would not accept his name. Despite the national belief that the United States is a country of continued growth and urbanization, some parts of the country have not necessarily experienced positive growth. In fact, much of the nation’s growth has occurred in the coastal regions while the geographic center of the US, known as the Great Plains, has struggled to ward off continued population loss attributable to out-migration. The release of the Census 2000 contributed to a series of headlines noting this remarkable population loss. North and South Dakota were featured in many of these articles, revealing that a great number of its counties have a lower population density today than in 1880, and that nearly 90% of North Dakota’s counties experienced a loss in population between 1990 and 2000. However, North and South Dakota and other Great Plains states have not always experienced population loss. Researchers are familiar with three main population surges between the mid-1800s through the end of the 20th century coinciding with the expansion of mining and ranching, the homestead era, and the “population turnaround.” These population increases add texture to what may at the surface seem a flat topic. In general, population change within the Great Plains is not a story of linear growth or decline, but, rather, a heterogeneous tale of booms and busts intertwined with structural forces, all played out at a local level. In this paper we assess the patterns of population change attributable to migration in North and South Dakota over the 20th century, and estimate the influence of “urbanization” factors—economic base, population characteristics, and spatial variables—on these patterns. Assessing population change within North and South Dakota is important to general discussions of urbanization and migration for two reasons. First, these states have not experienced the same level or type of population growth as observed for most other states within the US. The negative or no growth scenario is not widely studied by population researchers, yet such instances do in fact exist within the overall urbanization process. Second, although changes in population distribution are well documented, studies are mainly descriptive and period-specific, leaving questions of causality virtually unexplored and conceptions of population trajectories truncated. So, while we are aware of instances of population change, we are left wondering how change occurs, what form it takes, and whether its influences vary over time. Although the Great Plains consists of counties extending from Texas to Montana, this study focuses on counties within two states: North and South Dakota. These states were selected for three reasons. First, although many contesting definitions of the Great Plains exist, all include North and South Dakota. Second, a disproportionate amount of media coverage concerning population loss focuses on the Dakotas. This area has become the example of population loss within the US. Finally, these states are arguably representative of other areas within the larger Great Plains region experiencing population loss or stagnation. Therefore, the Dakotas serve as a reasonable laboratory to investigate long-term population change Using population, economic and agricultural census data, we employ growth curve modeling techniques to assess the structure of population change throughout the 20th century and estimate the impacts of the various “urbanization” indicators on this structure of change. The selected “urbanization” indicators reflect the county’s economic base (with specific focus on the agricultural industry), population characteristics, and spatial factors related to accessibility and amenities. Previous research has identified these indicators as contributing factors to population growth. According to the economic base perspective, people go where economic opportunities exist. Here, migration is viewed as a human capital investment, where people migrate from places with relatively low wage structures and little economic opportunity to places with higher wage structures and greater economic opportunity. Given the selectivity of the migration process, researchers have also identified population characteristics as a key player in growth or decline outcomes; certain people are more likely to migrate than others and selection criteria are rooted in population characteristics such as age and education. While a locality may begin with a normally distributed population, in terms of age and education, the migration process leaves the community with a hallowed out age structure and less educated population. Such characteristics reduce the likelihood of population growth and increase the chances of population decline. Yet, proponents of the spatial argument suggest that accessibility and proximity to a metropolitan center drives population growth. Within this perspective, population growth is oriented around the urban core, where places closer to the core experience greater growth compared to places further from the core. Accessibility operates similarly, such that accessible places are more likely to experience growth than isolated places. While each of these indicators has been identified in previous work, their impacts have not been empirically compared, nor has the idea that they are interrelated been systematically investigated. In this paper, each of the indicators’ relative contribution to county-level population change and the interrelationships among the indicators is assessed. The key feature of growth curve models is to relate variation in the pattern or trajectories of growth (or decline) to variation in the selected “urbanization” indicators. We begin by fitting time series models within counties for North and South Dakota from 1900 through 2000, and then examine how variation in the parameters representing the collection of county growth trajectories are related to the “urbanization” indicators. In addition, we describe the correlations between the patterns of growth or decline in selected indicators (e.g. agricultural production and education levels) and the population growth trajectories. One approach is to estimate population growth as a nonlinear function of time: Pit = boi + b1i (t) + b2i (t2), where i represents each county and t represents each year observed. At the same time, we estimate the variation in the function (or trajectory) as captured by the variation in the parameters across counties. For example, using this approach we relate the observed variance in b1 to initial economic and social characteristics of the county in 1900 or change in these characteristics. In this manner, we depict how the “main” effect of time is altered by the structural features of the counties. We will explore alternative forms of the time series to best depict the observed population change across counties from 1900 to 2000. One alternative is to fit simple spline models capturing expected boom or bust periods over the series. Preliminary analyses reveal both dramatic and subtle variation in county-level population change. Dramatic changes are noted at the beginning and end of the century, where some counties experienced a large drop between 1910 and 1920, and fewer counties experienced growth after 1970. Other counties experienced more subtle changes throughout the century, characterized by steady growth through 1940, followed by either a steady decline during the following years or a steady decline until 1980 where an increase in population is noted. The following stages of analysis include developing trajectories characterizing the major patterns of population change, assessing the impact of economic base, population characteristics, and spatial factors on these trajectories, and examining the interrelationships between these urbanization indicators.