Population Estimates and Projections-1989

Population Estimates
and Projqcflons
6h New Hanover Count
DCM COPY DCM COPY
i
lease do not remove!!!!!
Division of Coastal Management
[i V n lrVn H@_;tki ovaf Con PI.auinn,ng- De aift-nipton,l ii
�� ��o
The preparation of this document was financed in part
through a grant provided by the North Carolina Coastal
Management Program, through funds provided by the
Coastal Zone Management Act of 1972, as amended, which
is administered by the Office of Ocean and Coastal Re-
sources Management, National Oceanic and Atmospheric
Administration.
a

POPULATION ESTIMATES AND PROJECTIONS
IN NEW HANOVER COUNTY
I. Executive Summary .......................................................................... 1
II. Introduction...................................................................................... 2
A. Importance of Population Forecasts.................................................................. 2
B. Estimates vs. Projections...................................................................................... 2
C. Components of Population Change.................................................................. 2
D. General Considerations...................................................................................... 2
1. Area Size........................................................................................................ 2
2. Time Span...................................................................................................... 3
3. Complexity of Method.................................................................................. 3
4. Special Popuk tions........................................................................................ 3
S. Administrative Decisions................................................................................ 3
III. Methods Currently in Use.................................................................. 4
A. Extrapolation Method.......................................................................................... 4
B. Ratio Method.................................................................................................... 4
C. Cohort -Component Method................................................................................ 5
D. Economic Base Method...................................................................................... 5
E. land Use Method.................................................................................................. 6
IV. Methods Used for' Now Hanover County ........................................... 7
A. Estimates.............................................................................................................. 7
1. Regression (Ratio -Correlation) Method........................................................ 7
2. Administrative Records Method.................................................................... 7
B. Projections........................................................................................................... 8
1. Assumptions....................................................................................................8
a. Birth Assumptions...................................................................................... 8
b. Death Assumptions .............................................................................. 8
C. Migration Assumptions............................................................................ 8
2. Methodology .................................................................................................. 9
V. Other Potential Data Sources.......................................................... 10
A. Electrical Connectk)ns...................................................................................... 11
B. Telephone gook-ups.......................................................................................... 11
C. Voter Registration.............................................................................................. 13
D. Pubic School Enrollment.................................................................................... 13
VI. Recommendations........................................................................ 14
A. Utilize Alternate Data Sources.......................................................................... 14
B. Delay 1991 land Use Plan Update.................................................................... 14
C. Utilize Building Permit Reports............................................................................ I5
D. Improve Mobile Home Records........................................................................ 15
E. Explore land Use Methods.............:.......................................................:.......... 16
References........................................................................................... 17

I. EXECUTIVE SUMMARY
This paper examines population estimation and
projection methods available for New Hanover
County and makes recommendations on which
methods the County should use in future Land Use
Plan updates. Estimates can be defined as approxi-
mations of current or past population, while projec-
tions are forecasts of future population. Both
projections and estimates use U.S. Census data for
the base year population and build from that base.
The U.S. Census is taken every ten years.
Methods of predicting population vary in their
complexity and accuracy, but there is no one "best
method". Factors such as area size, time span, data
availability, technical expertise and use of data
dictate appropriate methodology. Methods com-
monly used include extrapolation, ratio, cohort -
component, economic base, and land use.
New Hanover County generally relies on the
State Data Center (SDC) for population estimates and
projections. Two methods are primarily used by the
SDC for estimating population: 1) the Regression
(Ratio -Correlation) method which utilizes federal
income tax returns, school enrollment, and automo-
bile registration, and 2) the Administrative Records
method, which uses birth and death records, federal
tax data and data on Medicare enrollees. For
projections, the SDC uses the Cohort -Component
method, which applies birth, death, and migration
rates to five-year age groups.
Several recommendations are made:
1) The County should collect data on residential
hook-ups of telephones and electricity.
2) The County should delay the upcoming 1991
Land Use Plan Update until after 1990 Census
data are available.
3) The County should utilize the Inspection Depart-
ment records of residential building permits
throughout the following decade(s) as a supple-
ment to the SDC figures.
4) The County's mobile home record -keeping
mechanisms should be improved. By modifying
the way data is collected and reported, the
mobile home data may be useful, not only for
the Tax and Inspections Departments, but also
as a planning tool.
5) The County should experiment with the Land -Use
method following the 1990 Census to gauge its
accuracy, and possibly use this method between
Census years.

II. INTRODUCTION
A. Importance of Population Forecasts
Knowledge of current and future population of
an area is vital to most sectors of society. Retail
businesses, for example, must know if a given
locality can support a new store. The location of a
factory may be contingent on a certain size
workforce. Utility companies must know where and
when to extend services. The future of many
businesses hinges on the level of population
growth.
Likewise, all levels of government need accurate
population estimates and projections in order to
efficiently provide adequate levels of services.
Engineering departments need population figures in
order to size and extend water and sewer lines.
other public services, such as police and fire
protection, social services, libraries, and hospitals
plan for future service levels based on population.
However, predicting the future is "an almost
impossible task, except in very general terms be-
cause of the amazing complexity of the social and
physical world." (Weeks, p. 344). Therefore,
predictions are not made; rather, demographers
estimate current and future populations based on
certain assumptions, primarily on the assumption
that the future will follow past trends.
B. Estimates vs. Projections
In this paper, population estimates refer to
current or past population. Population projections
refer to future population. In general, the basis for
both population estimates and projections are
Census counts from preceding years. Most com-
monly, these decennial year populations are "aged
forward" based on birth and survival rates. Migra-
tion is then approximated based on recent trends.
C. Components of Population Change
There are three components of population
change: Birth (fertility), death (mortality) and
migration. Births and deaths constitute natural
change and are relatively easy to forecast, based on
recent fertility and mortality rates. Migration is a
more complex issue. Relatively little is known about
why people migrate, and therefore migration is
much more difficult to measure than natural popula-
tion change. Net migration is the difference be-
tween in -migration and out -migration.
D. General Considerations
There are several factors which affect the
accuracy of population forecasting and/or the choice
of methodology. These include the size of the area,
time span, complexity of method, presence of
special populations, and the impact of administrative
decisions.
1. Area Size
In general, the smaller the area (population size
or geographic area), the greater the error that can be
expected. If the population is small, a change of a
few people can have a big impact on the rate of
population change.' Likewise if the geographic area
is small, then a small move may take a person into
or out of an area (Irwin, P. 5).
Migration, in particular, can have a large impact

on a small area. While immigration at the national
level is relatively stable, in some local areas migra-
tion becomes an enormous and unpredictable
component of population change. This may be the
result of gain/loss of a major employer, change in
immigration policies affecting certain nationalities,
historic patterns of settlement, and local economic
and regulatory differences. In metropolitan areas,
for example, suburban counties with lower tax rates
and fewer development regulations often experience
higher population growth than the central cities.
2. Time Span
The farther into the future population is forecast,
the less accurate it is likely to be. This is especially
true for smaller areas such as counties, municipali-
ties, and census tracts. Theoretically, population
estimates should be more accurate than projections,
since estimates refer to current or past population,
and can be based on known facts rather than
projected trends.
3. Complexity of Method
It is often assumed that the more complex the
method, the more accurate the projection or esti-
mate. This is probably true up to a point. (Weeks,
p.6.). In addition, more complex methods often
provide useful detail such as age, sex, and race
breakdowns, which may be necessary for certain
purposes. For example, planning for schools may
require an estimate of the number of children in
various age groups.
However, complexity does not guarantee
accuracy. Each new level of detail, i.e. age, sex, and
race, introduces a new element of uncertainty and
potential error. In addition, complex methods
require a larger data base, more time, and more
technical expertise to manipulate the numbers.
Simple methods should not be automatically re-
jected; rather, the method should be geared to the
needs of the project.
4. Special Populations
The presence of military installations, seasonal
housing, colleges and universities, prisons, and
other institutions can create problems in estimating
and projecting population. Gains and losses in
these special populations do not necessarily parallel
the local population. In addition, institutional
populations are often composed of persons within a
very narrow age range.
The usual method for handling this situation is
to forecast the non -institutional population, and
independently estimate the institutional population.
These two populations are then added together.
S. Administrative Decisions
Actions by governmental units can impact
population change. Decisions regarding the expan-
sion or contraction of military installations, colleges,
and other institutions are one example. There are
many other governmental actions which influence
the extent and pattern of local population growth.
(Weeks, p.8)
The federal government's role in housing and
highway construction, for example, hastened the
pattern of suburban growth over the past few
decades. Local government policy can also affect
population change. Residential construction can be
regulated through zoning restrictions, the issuance
of building permits and construction of water and
sewer lines. These factors can encourage popula-
tion growth in certain areas, or severely restrict it.
3

III. METHODS CURRENTLY IN USE
A. Extrapolation Method
The simplest approach to projecting population
is extrapolation. Extrapolation methods use past
growth rates and extend them into the future. These
methods require the least amount of data of all the
techniques.
One type of extrapolation assumes arithmetic
growth. For example, if an area grew by 1,000
persons over ten years, an additional 1,000 persons
per ten-year period could be projected. A second
type is geometric extrapolation, which assumes the
rate (percent) of population change will be repeated
in succeeding decades (Raymondo, p. 38).
Figure 1 uses 1950 and 1960 data for New
Hanover County to extrapolate 1970 and 1980
population. In this case, the geometric method was
more accurate than the arithmetic method and was
fairly close for the actual 1970 population. How-
ever, the 1980 estimates were considerably less than
actual population, due to large population growth in
the unincorporated County during the 1970s.
Figure 1: Extrapolation Method
Actual
Population
Increase 1950-60
Population Estimate
Actual
JPercent
Arithm.
Geom.
1950 63,273
1960 71,742
8,469
13.4%
1970 82,996
80,211
81,355
1980 103,471
88,680
92,257
Thus, the extrapolation method is a fairly crude
indicator*of population growth. For areas with a
relatively stable population growth, this method may
produce accurate results over a short time span:
However, extrapolation may not be appropriate for
forecasts far into the future and in times of turbulent
population change.
B. Ratio Method
The ratio method expresses an area's population
as a ratio or proportion of a larger "parent" area's
population for which a projection or estimate
already exists. The proportion can be stable or
changing. For example, if a county has 5% of the
state's population, the "stable ratio" assumes that in
ten years, the county will still have 5% of the state's
population. The "changing ratio" approach takes the
change in the rate of the county's proportion into
account (Raymondo, p. 38).
Figure 2 estimates 1970 and 1980 population of
New Hanover County based on 1950 and 1960 ratios
and assuming the State's population is known. As
this example shows, the ratio method was fairly
close for the 1970 estimate, but the County's acceler-
ated growth was not reflected in the 1980 estimate.
The stable ratio assumed that New Hanover County
would remain at 1.57% of North Carolina's popula-
tion, while the changing ratio assumed 1.58% for
1970 and 1.59% for 1980. The actual ratios were
1.63% for 1970 and 1.76% for 1980.
Thus, like the extrapolation method, the ratio
method is most appropriate for short-range projec-
tions, and for areas with a stable pattern of change.
4

Figure 2: Ratio Method
Actual Po ulation
NH Co. as
% of NC
Stable .
Ratio
Changing
Ratio
NH Co.
NC
1950
63,273
4,061,929
1.56%
1960
71,742
4,556,155
1.57%
1970
82,996
5,084,411
79,825
80,334
1980.
103,471
5,881,766
92,344
93,520
Furthermore, with both methods, the choice of base
year(s) will impact the results. Care must be taken
to prevent increasing ratios from becoming too large
and decreasing ratios from declining too rapidly.
One solution is to assume the ratios will reach
stability at a specified future date and adjust the
projected ratios from that date (Irwin, p.14).
The ratio and extrapolation methods are not
generally used as a sole technique for population
estimates and projections. Rather, they are used in
conjunction with other methods. The ratio method,
for example, is often used to apportion population
among municipalities and unincorporated areas
within a county whose population has been pro-
jected using another method.
C. Cohort -Component Method
The cohort -component method projects popula-
tion by age groups or "cohorts". Most commonly,
five-year age groups are aged forward in five or ten-
year increments. The persons aged 15-19 years old
in 1980, for example, belong to the 20-24 year -old
cohort in 1985, and the 25-29 year -old cohort in
1990. The cohorts are usually divided by sex and
sometimes by race as well.
The "components" of population change -
births, (fertility), deaths (mortality), and migration -
are applied to the cohorts to forecast future popula-
tion. Birthrates are available from state vital statistics
offices. The general fertility rate (number of births
per 1,000 women aged 1544) or age -specific fertility
rates can be used. These rates are then applied to
the number of women in child-bearing years (usu-
ally considered to be 1544) to get the number of
births for the projection period. Mortality is esti-
mated using survival rates, which are available from.
the National Center for Health Statistics, or state vital
statistics offices. Assume, for example, there are
3,500 persons aged 15-19 in an area in 1980, and the
ten-year survival rate for this cohort is 0.95 (95%).
In 1990, there will be 3,325 persons in this cohort,
which will then be aged 25-29.
Net migration is forecast based on estimates of
past migration. Generally, the most recent two
decennial censuses are compared. Survival rates are
applied by cohort to the earlier census to compute
the number of people who would be counted in the
most recent census assuming zero migration. The
difference between this estimated count and the
actual census count represents net migration for the
base decade. This migration rate can then be
assumed to continue into the projection year, or may
be adjusted to take other factors into account.
The cohort -component method is useful for
many planning functions, since it can yield detail by
age, sex, and race. Planners of housing, schools,
medical facilities aimed at certain ages (i.e. pediat-
rics, cardiology), and retirement services, for ex-
ample, require accurate projections by age and may
justify the use of the cohort -component method.
However, the high level of detail requires more data
input and more calculation - a computer is essential.
Thus, the use of the cohort -component method for
population projections is a major undertaking
(Irwin, p. 23).
D. Economic Base Method
Economic base, or "econometric" methods base
population projections on projected economic
conditiops. These methods are most often used to
forecast net migration, with natural population
change (births and deaths) projected separately.
One type of econometric model uses a regres-
sion equation to predict net migration. Several
models have been advanced, using such predictor
variables as income, college enrollment, unemploy-
ment rate, labor supply, population density, occupa-
tion, race, labor force participation, and -distance
from other cities (Irwin, p. 26-27). While causal
relationships between employment and migration
have not been established, one author concluded
there exists "a series of feedbacks or mutual adjust-
ments between migration and the demand for labor"
(Lowry, as quoted in Irwin, p. 25).
Another method projects.an area's future
employment needs by.asking major local employers
about expansion plans, applying state or national
employment projections to the local level, and/or
making assumptions about employment growth.
Available labor force is then projected (assuming
zero migration) by applying estimated labor force
participation and unemployment rates to the pro-
5

jected working age population. If the projected
labor force is smaller than future employment needs,
immigration is assumed (Raymondo, p.38).
Econometric models are not widely used for
local population projections because local employ-
ment growth does not always result in local popula-
tion growth (Raymondo, p. 39). An increase in jobs
in New Hanover County, for example, may result in
population growth in neighboring Pender or Brun-
swick Counties.
E. Land Use Method
An examination of land use can reveal patterns
of housing (and thus population) growth. One
approach measures the "saturation" or maximum
housing density an area can accommodate. This
method uses zoning restrictions (minimum lot size)
to determine housing count assuming full buildout.
Multiplying this by estimated average persons per
household yields a population projection (Irwin, p.
27).
This method can also be used to estimate
current population. A housing count can be esti-
mated either by aerial photos, "windshield surveys,"
or an examination of tax records, which list land use
for each parcel. Estimated population is then
determined by multiplying the housing count by
persons per household.
These land use methods are well suited for
estimating and projecting population for small areas
such as census tracts (Irwin, p. 27). The New
Hanover County Planning Department, in fact, uses
land use methods both to estimatepopulation and
housing (windshield survey method) and to project
population (saturation method)jorsmall areas.
F1

IV. METHODS USED FOR NEW HANOVER COUNTY
New Hanover County generally relies on the
North Carolina State Data Center (SDC) for County
population estimates and projections. Two methods
are primarily used by the SDC for estimating popula-
tion: 1) the Regression (Ratio -Correlation) method,
and 2) the Administrative Records method. For
projections, the SDC uses the Cohort -Component
method.
A. Estimates
Population estimates for North Carolina counties
are prepared through the Federal -State Cooperative
Program for Local Population Estimates. Two
methods are used. The methods are then averaged
and forced to sum to the state estimate. Both
methods independently estimate institutional
population. Figure 3 shows the 1987 State Data
Center population estimates for New Hanover
County and North Carolina. -
Figure 3: Population Estimates 1987
1. Regression (Ratio -Correlation) Method
This method uses multiple regression to deter-
mine what factors best predicted 1980 Census non -
group quarters population for the County from the
1970 figures. The factors - federal income tax
returns, school enrollments grade 1-8, and automo-
bile registrations - are then used to estimate current
non -group quarters population. The resulting value
is then combined with independent estimates of
military barracks, college dorms, and other institu-
tions.
2. Administrative Records Method
The Administrative Records method uses birth
and death statistics to measure natural change in
population since the most recent Census. Migration
is measured using federal tax data in the base
(Census) and estimate years to determine the
number of persons whose county of residence has
New Hanover
County
North Carolina
July 1987 estimate
116,337
6,411,810
April 1980 Census
103,471
5,880,415
Difference 1980-87
12,866 (12.4%)
531,395 (9.00/0)
Births
10,382
627,605
Deaths
6,727
370,157
Migration Estimate
9,211 (8.91/6)
273,947 (4.7%)
Source: North Carolina State Data Center, Office of State Budget and Management, 1988.
7

Figure 4: Population Projections 1988.2010
1980
Census
1988
Pm'
1990
Proi
2000
Pro'
I 2010
Pro
New Hanover County
White Male
38,791
44,619
45,887
52,664
58,595
White Female
41,562
48,332
49,746
57,372
64,063
Non -White Male
10,573
11,378
11,557
12,571
13,503
Non _White Female
12,545
13,783
14,028
15,366
16,524
Total
103,471
118,112
121,218
137,973
152,685
North Carolina
White Male
2,176,227
2,388,488
2,433,097
2,663,861
2,846,959
White Female
2,280,408
2,526,222
2,577,863
2,829,015
3,024,954
Non -White Male
678,539
733,794
745,940
811,457
865,213
Non -White Female
745,241
836,604
856,491
965,415
1,038,853
Total
5,880,415
6,485,108
6,613,391
7,260,748
7,775,979
Source: North Carolina State Data Center, Office of State Budget and Management,1988.
Note: Includes Wilmington and Carolina, Wrightsville, and Kure Beaches along with unincorporated County.
changed. In addition, data on Medicare enrollees is
used to estimate population aged 65 and over.
These estimates are then combined with estimates of
group quarters populations.
B. Projections
The Cohort -Component method is used to
project population of North Carolina counties. Since
projections are more likely to be accurate for large
areas, county projections are forced to sum to state
projections. Following is a discussion of assump-
tions and methodology of the Cohort -Component
technique as used by the S.D.C. for projecting
population of North Carolina counties. Population
projections for the County and State are shown in
Figure 4.
1. Assumptions
As noted in the introduction, demographers do
not predict population per se; rather, projections are
based on certain assumptions. If the assumptions
are correct, the projections will be accurate.
a. Birth Assumptions
North Carolina's fertility rates (children born to
the average woman during her lifetime) are less than
the U.S. average. White fertility is expected to
increase slightly, while non -white fertility will
decline. Figure 5 compares 1985 and 2010 fertility
rates. Both white and non -white births are expected
to peak during the 1980s and early 1990s, as the
"baby boom" cohorts move through prime child-
bearing years.
Figure 5: Fertility Rates
North Carolina
U.S.
1985
2010
1985
2010
White
Non -white
1.56
1.87
1.57
1.75
1.75
2.17
1.79
2.04
b. Death Assumptions
North Carolina's death rates are slightly higher
than the U.S. overall. Over the next century the
State's death rates are expected to converge toward
national rates, reaching equality by 2080.
Life expectancy at birth is projected to increase
three years from 1988-2010. However, the number
of deaths will increase, as the increasing elderly
population dominates this improvement in life
expectancy.
c. Migration Assumptions
Net migration is projected to follow recent
experience. During 1980-2010, North Carolina
should experience net migration of about 40,000 per
year as follows:
Figure 6: Net Migration
Net Migration
1980-1990
400,905
1990-2000
404,202
2000-2010
395,891
8

2. Methodology
The Cohort -Component method is applied to
State projections essentially as described in Section
III C (p. 5). The projections are divided by age, sex,
and race. Births are estimated using age- and race -
specific birth rates. Deaths'are estimated by apply-
ing survival rates to the projection decade. Migra-
tion is forecast based on trends established in the
previous two decennial censuses.
Projections are modified in some counties based
on the presence of such facilities as military installa-
tions, prisons, and major universities. However,
New Hanover County is not among the counties in
which the projections are modified. The county
projections are forced to sum to State age, sex, and
race totals.
9

V. OTHER POTENTIAL DATA SOURCES
The most commonly used data source for
population estimates and projections is the U.S.
Census. However, since the Census is only taken
every ten years, demographers often examine other
data sources. Sources used by the NC State Data
Center, for example, include federal income tax
returns, school enrollments, automobile registrations,
birth and death records, and Medicare records.
Several sources of data were examined in the
course of preparing this report: electrical connec-
130
120
w
110
100
.-• 90
so
L c 70
�G. 60
s0
40
30
20
10
0
G Census
(population)
octud
— — — projected
[ions (Carolina Power and Light), telephone hook-
ups (Southern Bell), voter registration, and public
school enrollment. Since data are gathered yearly,
these sources may provide an alternative method of
estimating population and recognizing trends.
Historical data were available for these agencies and
are compared to Census data and estimates/projec-
tions in Figure 7. Following is a discussion of these
data sources and their potential use for projections
and estimates. In addition, some agencies also
project population, as discussed below.
Figure 7: Population Data Sources
Census vs. Other Sources
•.o
i
1950 1960 1970 1980 1990
+ CP&L o Southern Bell - e Voter Registration x School Enrollment
(households) (households) (persons) (students)
10

A. Electrical Connections
Electricity is supplied to New Hanover County
by Carolina Power and Light (CP&L). Figure 8
shows the number of residential customers from
1980-1988. The data include the number of house-
holds billed at year end. Since there are always
some households in the process of connecting/
disconnecting, the total number of customers is
probably slightly higher.
Figure 8: Carolina Power & Light
Residential Customers 1980-88
1980
39,156
1981
40,474
1982
41,868
1983
43,127
1984
45,386
1985
47,717
1986
49,596
1987
51,294
1988
52,686
Source: Carolina Power & Light, 1989.
CP&L uses projections to a certain extent in .
planning for future electrical service. However, their
primary planning is through examining load infor-
mation (demand for electricity) at each substation.
Load is measured at peak times - the hottest days of
summer and coldest days of winter. Facilities are
planned based mainly on these load trends and
substation capacity. Trends in housing growth are
.factored in each year. However, for projecting
demand for electrical services, CP&L has found the
substation data more accurate than housing esti-
mates.
B. Telephone Hook-ups
Southern Bell provides telephone service to New
Hanover County. Historical data on telephone
hook-ups are shown in Figure 9. Southern Bell's
service districts are divided into "wire centers".
There are six wire centers serving New Hanover
County. Two centers overlap the Pender-New
Hanover County boundary. The Scotts Hill and
Castle Hayne Wire Centers provide approximately
80% and 50%, respectively, of their service in New
Hanover County (See Map 1).
The company uses several methods to forecast
future housing growth and therefore to predict
where new phone hook-ups will be required. In
previous years, comprehensive housing counts were
done approximately every five years by windshield
surveys. However, due to budget cutbacks, these
surveys are no longer done.
Figure 9: Southern Bell
Residential Customers 1976-89
1976
31,268
1977
32,412
1978
32,595
1979
33,923
1980
35,534
1981
36,433
1982
37,386
1983
37,660
1984
38,639
1985
40,143
1986
41,466
1987
43,433
1988
45,266
1989
47,235
Source: Southern Bell, 1989.
Note: Includes 80% of Scotts Hill and
50%of Castle Hayne Wire Centers.
Southern Bell currently relies on a computer
program called "American Profiles", which uses
Census data. The wire center boundaries are
overlaid on Census maps, and the computer pro-
gram generates Census data for the wire centers.
This program is supplemented with information
about residential construction. Residential building
permits are recorded for every county to see where
and how rapidly growth is occurring. In addition,
development in new subdivisions is monitored on a
quarterly basis, by means of windshield surveys.
Trends in telephone hook-ups may be an
indicator that could be used to estimate current
population. However, there are limitations that
should be recognized. The main problem is that
hook-ups are recorded in terms of wire centers,
rather than municipalities and counties. Two of the
wire centers in New Hanover County are located
partially in Pender County.
Another problem with using telephone hook-
ups to estimate population is that the percentage of
households having phones differs in various areas.
For example, a beach town, with its seasonal
population, may have fewer hook-ups per house-
hold. Lower income areas will likewise have fewer
telephones.
Finally, there would be a problem in recogniz-
ing trends. Ideally, a historic correlation could be
made linking population (households) with tele-
phone hook-ups. However, it should be recognized
that there may be an increase in the percentage of
households having phones through time. Also,
more and more households have more than one
phone line, whether for a business, children's
phone, computer modem, or fax machine.
11

i�
1
q
L
---- SCMA
�.Iiq.
1.7,
.yam ��-_'_�: +�5���,`,. .a. •
CRBH
­0 N.C.
Q,
Map 1: Southern Bell
Wire Centers
CSITY '
Castle Hayne
SCMA
Scotts Hill Main
WLWI
Winter Park
WGVL
Wrightsville Beach
WLFO
Fourth Street
CRBH
Carolina Beach
12

C. Voter Registration
The New Hanover County Board of Elections
keeps records on registered voters for the County.
Historic voter registration records, available from
1943, are shown in Figure 10.
Figure 10: Voter Registration
1943-89 . _
1943
8,952
1944
14,795
1946
9,596
1947
7,793
1953
17,587
1956
33,022
1964
30,742
1965
28,061
1966
29,989
1968
31,238
1969
30,958
1970
31,458
1971
31,731
1972
37,410
1973
35,596
1974
36,644
1975
37,399
1976
42,392
1977
39,111
1978
41,110
1979
43,380
1980
46,527
1981
43,991
1982
45,725
1983
48,712
1984
58,269
1985
52,808
1986
54,321
1987
54,662
1988
61,892
1989
54,345
Source: New Hanover County Board
of Elections, 1989.
There are several factors that affect the reliability
of voter registration as a source for estimating
population. First, females are over -represented in
voter registration rolls, since they are more likely to
register (and vote). Second, various circumstances
affect voter registration - campaigns to "get out the
vote;" emotional elections, either at the local or
national level; and the tendency for people to
register in presidential election years. Third, the
registration rolls are "purged" every four years, after
the presidential election. Any voter who has not
voted in four years is dropped if he does not
respond to a letter. Finally, the 26th Amendment to
the constitution gave 18 year -olds the right to vote.
This legislation, enacted in 1970 and in effect in
1972, should be considered when establishing
trends.
D. Public School Enrollment
Enrollments in the public school system may
provide an indicator of population change. Data
from the New Hanover County Board of Education
are shown in Figure 11. These data show "member-
ship", or the number of pupils in school on the last
day of class. Due to a fire at the school board
building in 1971, enrollment figures are not available
before the 1971-72 school year.
Like the previously discussed data sources, there
are limitations to the use of public school enroll-
ments to estimate population. First, birthrates in
previous years affect enrollment. The "baby boom,"
"baby bust," and "baby boomlet" create a ripple
effect through time as these generations age. Sec-
ond, the trend through time has been fewer children
per family. Third, there has been an increase in
retirees in the County. Since they generally don't
have school age children, this immigration would
not affect school enrollments. Fourth, this measure
overlooks the impact of private schools, both in
New Hanover County and outside the County.
Finally, North Carolina introduced public school kin-
dergarten in 1969. The program became fully
funded in the 1976-77 school year, making kinder-
garten available for every child in the state.
Figure 11: Public School Enrollment
1971-88
1971-72
18,936
1972-73
19,018
1973-74
18,951
1974-75
19,467
1975-76
19,697
1976-77
20,550
1977-78
20,435
1978-79
20,348
1979-80
20,135
1980-81
19,775
1981-82
19,277
1982-83
19,099
1983-84
18,813
1984-85
18,566
1985-86
18,720
1986-87
18,813
1987-88
18,621
Source: New Hanover County Board
of Education, 1989•
13

VI. RECOMMENDATIONS
In the absence of frequent Census counts, it is
difficult to gauge the accuracy of population esti-
mates and projections. While the SDC figures are
based on sound methodology, rapid growth that has
taken place along the State's coast may not be
reflected in these estimates and projections. In
addition, the population estimates are often two to
three years old. The following recommendations
can be made to improve population estimates and
projections for the County.
A. Utilize Alternate Data Sources
It was hoped that the data sources discussed in
the previous section would provide an alternate
method of estimating population and recognizing
trends. However, these data sources have limita-
tions.
The primary problem with these data sources is
the inability to discern a correlation. Ideally, a
multiple regression equation could be used to see
which data sources, i.e. voter registration, telephone
hook-ups, etc., best predict actual population. Two
of the sources, however, CP&L connections and
Southern Bell hook-ups, are available only from
1980 and 1976, respectively. Since the most recent
Census was done in 1980, a trend cannot be estab-
lished. It may be useful to continue to collect data
from CP&L and Southern Bell through 1990, and
then compare them to Census data:
an increasing trend, like population growth: The
difficulty with this source is the cycle of increased
registration during presidential election years, -
followed by a purge the following year. This "zig-
zag" pattern would make it difficult to estimate
population in any given year.
While it is felt that records from CP&L and
Southern Bell have potential use for population
estimates, more data needs to be gathered to
establish trends. Figure 12 (page 15) compares these
utility connections to 1960-90 occupied housing
counts from the Census. The 1990 Census housing
count is projected, and it appears that this projection
may be low, based on telephone and electrical
connections. It is recommended that the County
continue to use estimates from the SDC for the short
term, and continue to collect data on utility (tele-
phone and electrical) connections.
B. Delay 1991 Land Use Plan Update
The Coastal Area Management Act of 1974
requires that each local government develop a Land
Use Plan and update it every five years. The
County's Land Use Plan (LUP) which provides the
basis for long-range planning, is scheduled to be
updated in 1991. Preliminary technical reports
precede the LUP process and take approximately
one to two years to complete.
One major component of the LUP process is a
A third source, school enrollment, is available technical report on population. It is recommended
from the 1971-72 school year. However, the trend that this report be delayed until 1990 Census data
has been a steady or declining enrollment, in are available. Census data would provide a reliable
contrast with general population growth. Finally, base for population estimates and projections, which
voter registration, collected since 1943, has shown are vital for the development of the LUP update.
14

54
52
50
48
46
44
42
40
38
36
34
32
30
28
26
24
22
20
Figure 12: Population Data Sources
Census Housing Data vs. Other Sources
.+ CP&L O S Bell A Occ. Housing
Delaying this technical report may result in a delay
in the LUP Update of up to one year. However, it is
felt that an improvement in the accuracy of the data
will offset any inconvenience caused by this delay.
C. Utilize Building Permit Reports
The County Inspection Department keeps
excellent records of residential construction activity,
in the form of building permits, for Wilmington and
unincorporated New Hanover County. The towns of
Wrightsville Beach, Carolina Beach, and Kure Beach
also keep records of building permits issued. These
records are utilized by the Planning Department in
the annual update of the Construction Activity
Report. While this report focuses on value of new
construction, figures are also available on the
number of permits issued.
It is recommended that records be collected
each year on: 1) number of permits issued for new,
residential construction, and 2) number of permits,
issued for demolitions. For multi -family construc-
tion, data should be collected by number of dwell-
ing units.
This information, coupled with 1990 Census data
and mobile home data (see below) should yield a
fairly accurate and timely housing count. House-
hold size estimates (from the 1990 Census) can then
be applied to determine a population estimate.
— — projected
D. Improve Mobile Home Records
In addition to issuing permits for construction,
the County Inspection Department also inspects and
issues permits for mobile homes prior to occupancy.
Since mobile homes provide a major source of
housing in New Hanover County, accurate mobile
home records are invaluable in estimating popula-
tion.
.. However, there are problems with the data as
they are currently collected. For example, a recent
update of the sewer plan by the firm Talbert, Cox
and Associates revealed that the County's tax records
do not account for mobile homes adequately. The
following recommendations, if implemented, will
make the mobile home data useful for population
estimates. These suggestions apply to the monthly
permit report (INS 155) and will require'some
alteration on the form used by the Inspection
Department.
1. Differentiate between mobile homes that are
a) new, b) moved from another location in the
County, and c) moved from outside the County. A
mobile home at a new location must be inspected
even if it is moved within the County. If these intra-
county moves were separated, the mobile homes
that are new or new to the County could be added
to the housing count each year. One remaining
problem would be accounting for mobile homes
which are moved out of the County.
15

2. Differentiate between mobile homes intended
for residential versus office/commercial uses.
3. List the jurisdiction (municipality or unincorpo-
rated County) in which the mobile home is located.
These changes would make the mobile home
data much more useful for planning purposes.
Combining 1990 Census housing counts with yearly
counts of new residential construction and new
mobile homes should result in accurate housing
counts. These estimates could then be used to
establish trends and make projections.
E. Explore Land Use Method
Another potential data source is the County Tax
Department. Computerized records are kept for
each parcel of land in the County. These records
include a code that denotes land use, i.e. 01 = single
family residential, 02 = duplex, 10 = mobile home,
etc. Using these tax records, supplemented with
field research, land use is mapped every five years
during the update of the Land Use Plan.
New Hanover County, in conjunction with the
City of Wilmington, has recently acquired a geo-
graphic information system that will allow the tax
data to be linked via computer to parcels of land on
a map. The system, known as MAPS (Management
and Planning System), will result in a much more
rapid mapping process during the next LUP update,
since these maps were previously compiled by hand.
However, field research will still be necessary, since
the land use code entries are not always up to date.
It is recommended that residential uses on the
resulting land use maps be tallied and compared
with Census counts. The Land Use Map update
should be compiled as dose as possible to the
Census count date, April 1, 1990. If the land use
method proves to be an accurate method of estimat-
ing population, it can be employed between Census
years.
Two additional recommendations are made con-
cerning the implementation of the Land Use Method.
First, the Inspection Department should tie all
inspections to the parcel's tax map number. This
will make it easier for the Tax Department to update
their records. Second, the Tax Department should
establish a procedure to regularly update the land
use code entries. These recommendations should
result in a more accurate Land Use Map.
16

REFERENCES
Carolina Power and Light, residential customer data,
1989.
Irwin, Richard, Guide for Local Area Population -Pro-
jections, U.S. Bureau of the Census Technical
Paper 39, Washington DC: Government Printing
Office,.1977.
Lowry, Ira S., "Migration and Metropolitan Growth:
Two Analytical Models," San Francisco, Califor-
nia: Chandler Publishing Company, 1966, as
quoted in Irwin, 1977.
New Hanover County Board of Education, public
school enrollment data, 1989.
New Hanover County Board of Elections, voter
registration data, 1989.
New Hanover County Planning Department, Popula-
tion Study of Netu Hanover County, Wilmington,
August 1985.
North Carolina Office of State Budget and
Management, North Carolina Municipal Popula-
tion, 1987, Raliegh, October 1988.
North Carolina Office of State Budget and
Management, North Carolina Population Projec-
tions: 1988-2010, Raleigh, July 1988.
North Carolina Office of State Budget and
Management, North Carolina Population Projec-
tions Assumptions and Methods, Raleigh, 1989.
North Carolina Office of State Budget and
Management, North Carolina Population Projec-
tions: Detailed Methodology, Raleigh, 1988.
North Carolina Office of State Budget and
Management, Provisional Estimates of the
Population of North Carolina Counties, July 1,
1987, Raleigh, July 1988.
Raymondo, James C., "How to Choose a Projection
Technique, American Demographics, February
1989, Vol. 11, No. 2, pp. 38-39.
Southern Bell, residential customer data, 1989.
Weeks, John R., Population: An Introduction to
Concepts and Issues, Belmont, California:
Woodsworth Publishing Company,1981.
17