The growth model for Hall County is an analysis tool and not a population forecast. We are often asked for a Hall County population forecast. A simple linear trend extrapolation (LINE) based on 1990 and 2000 decennial census data is summarized in the following table.
We acknowledge that the above data and projections do not take the Hispanic, or other, undercounts into consideration. The following table adjusts the Hispanic totals by the arguably conservative factor of 1.5. The mathematics involved in adjusting for the Hispanic undercount are straight forward and we encourage readers to explore variations if it is believed that 1.5 is too small an adjustment factor. See Hispanic Undercount for additional information.
State and Local Population Projections Methodology and Analysis by Stanley K. Smith, University of Florida, Jeff Tayman, San Diego Association of Governments, and David A. Swanson, Dean of the Helsinki School of Economics and Business Administration, and published by Kluwer Academic/Plenum Publishers in 2001 as part of the Plenum Series on Demographic Methods and Population Analysis. This book is a comprehensive review and analysis of small population projection methodologies.
Dr. William H. Frey published Three America's: The Rising Significance of Regions in the Fall 2002 edition of the Journal of the American Planning Association. The linked PDF file is a pre-publication version without maps, tables, and graphs.
We use these two sources in the following statistical and growth analysis of the population forecast. They are both current and authoritative. We welcome any comparable quality references on internal migration patterns, population projection analysis, Hispanic social factors, etc. that are pertinent to this work.
The 2002 Georgia County Guide
by the College of Agricultural and Environmental Sciences,
College of Family and Consumer Sciences,
The University of Georgia, Athens, Georgia.
This publication consolidates a wealth of information from
a number of data sources.
As with all data of this sort,
it is easy to be mislead by the precision and orderliness of
the data to believe that it is 100% accurate.
Please be aware that there are sampling and reporting errors present in this data.
On the other hand, it is, as the title implies, an
Our analysis is divided into two sections, a statistical analysis followed by a review of the social factors that are driving Atlanta regional growth.
All of the references in this section are from State and Local Population Projections Methodology and Analysis cited above.
There are three intertwined points of discussion that can be approached from a statistical standpoint. The first point is the choice of the LINE method for the above population projection. A survey of population projection methodology studies indicates that the LINE method predicts future total population counts as well as, or better than, other demographic methodologies such as polynomial regressions, cohort-component methods, or structural models (Chapter 13, Forecast Accuracy and Bias).
While the accuracy of the total population count is surprisingly good with LINE, the authors are clear about the relative lack of usefulness of trend extrapolation projections in the context of a planning process.
A total population count forecast was requested and we utilized the LINE method for generating that forecast. Our Hall County growth model is provided as a rudimentary cohort-component/structural model population growth study tool. Time and cost considerations prohibit a full cohort-component or structural analysis at this point in time.
The second statistical point of discussion is the pattern of significantly high (or low) population growth rates to change over time to an average growth rate. This point is intwined in the first because it is precisely this statistical tendency that allows LINE projections to more accurately project population counts than other methods. It is easy to get caught up in the assumption that growth will continue at or near its current rate in a high growth area and to use a projection methodology that captures that assumption. Statistically over a long horizon such as 30 years this assumption rarely plays out.
The LINE method has a slight tendency to underestimate the population, particularly for a high base growth rate (pgs. 311, 317, 318). However, the ten year base provided by 1990 and 2000 captures the highest growth rate period for Hall County and both building permit and water meter data indicates a decline in the rate of growth after 1999. We avoid applying a bias correction to the above projections for this reason.
The third and final statistical point is the estimate of the mean absolute percent error for LINE projections for our population characteristics. As noted above, the LINE method exhibits a mean absolute percent error of slightly over 40% for population projections over a 30 year horizon with a ten year base for populations under 1 million and a ten year growth rate over 25% during the base period (pg. 322). However, that number is based on a study of the LINE method applied to state population projections. Another approach is to note that LINE projection errors tend to increase linearly up to 35 year horizons (pg. 321) and the 10 year horizon mean absolute percent error is 9.3 when county population size is factored in (pg. 317) and 13.9 when county base growth rate is factored in (pg. 318) which, in combination, lends credence to a 30 year mean absolute percent error estimate of approximately 40%.
With births less deaths yielding a natural growth rate of less than one percent for the United States as a whole, for a county, state, or region to display the type of high growth rates found in the Atlanta region during the 1990s is significant. Our Atlanta region growth comes from a combination of both internal migration (from one region of the United States to another) and international migration, with the dominant factor being internal migration. This internal migration is important to understand because it affects the whole of our country. Planners need to know why people decide to leave one part of the country and move to another.
Dr. William H. Frey has extensively studied our country's internal migratory patterns during the 1990s and published his findings in the Fall 2002 edition of the Journal of the American Planning Association. in the article titled Three America's: The Rising Significance of Regions.
The main point of the article is that the core population movement within the United States during the 1990s was a white flight from diverse urbanized regions, i.e. the Melting Pot, to homogenous suburbs and small towns, i.e. the New Sunbelt.
The next question is whether Hall County provides that homogenous lifestyle sought by the these regional white flight migrants. Using census data listed in The 2002 Georgia County Guide, we find that of the 8 Georgia counties with growth rates over 7% from 2000 to 2001, none had a 2000 Hispanic population over 6%. In contrast, Hall County had a 4.6% growth rate and a 19.6% Hispanic population.
Referring to Gainesville-Hall County single family home building permit data we find permits peaked in 1999. This data is confirmed by the number of water meters sold by the City of Gainesville Public Utilities.
Historically, referring again to undercounted census data listed in The 2002 Georgia County Guide, we find that during the 1990s Hall County grew by 22,684 Hispanics and by 21,159 non-Hispanics. This is a 498% increase in the Hispanic community and a 23% increase in the non-Hispanic community.
In summary, Hall County growth during the 1990s
was dominated by an Hispanic migration
and the county's resultant culturally diverse makeup
may divert some of the white bedroom community
drive to less culturally diverse neighbors.
We continue to research this important issue. Currently we are interviewing community leaders and developers to better understand the recent decline in housing. We would appreciate hearing any anecdotal or statistical evidence that bears on Hall County growth.
June 23, 2003
Nancy G. Harden; firstname.lastname@example.org
include "../mm/footer.inc" ?>