Population Projections Description
The current set of population projections for the state and its counties was released by the North Carolina Office of State Budget and Management in October of 2016 and updates the release of September 2015. State and county total population projections are made using times series trends (exponential smoothing and ARIMA models).
Like the previous model, this model considers both the trend growth of 1990-2010, the annual non-institutional growth from July 1, 2010 through July 1, 2015 derived from the set of July 2000-2009 Smoothed County estimates (intercensal estimates), the July 1, 2010-2014 revised population estimates, as well as the 2000 and 2010 Censuses. For the current set of projections, the assumption is that migration (and, hence, growth) for July 1, 2016 through July 1, 2036 is also a function of the "trend" growth based on the 1990-2010 base decades, and the average annual non-institutional growth through July 1, 2015 derived from the set of July 2015 County Estimates.
For the current set of projections, an exponential smoothing or ARIMA model was selected that most accurately projected the 2010 Census and subsequent (2010-2015) county estimates while maintaining a low MAPE for each county using 1990-2015 data, and the county totals were summed to obtain projections for the entire state.
The most fundamental basis for these projections are population values for North Carolina and its counties from the Census Bureau's 2010 Census of Population. Since the last projection series, there have been a few minor corrections based on Census Bureau boundary changes. There was also a significant change of group quarters population from Granville County to Durham County which was the result of a Census CQR correction.
These projections use single year of age totals (0-99) and one composite age group (100+) by male and female from the 2010 Census for the data for the 2nd year of the base decade. The Census Bureau's modified race and sex file for North Carolina was used to assign the race categories for the 2010 base.
The "aging forward" of this 2010 population base strongly affects these age, race, and sex specific projections. This effect causes projected births to rise more slowly than projected population, in spite of fertility rates, which are assumed to remain constant. It also causes projected deaths to increase more rapidly.
Population values for North Carolina and its counties from the Census Bureau's 2000 Census of Population are very important as well. The actual data came from the Census Bureau's MARS (Modified Age, Race, and Sex) file. The most critical use of these values is in the determination of detailed migration trends for the 2000-2010 time period.
- Birth Assumptions
Since the early 1990s, North Carolina's fertility rates have changed significantly. In recent years the fertility rate declined significantly with the Recession, and has now leveled off or increased slightly. It was assumed that fertility rates for all age and race groups would be constant from 2015 through 2036. To obtain a more stable set of future rates, the fertility rates for 2013, the fertility rates for 2014, and the fertility rates for 2015 were averaged. Rates for each county are adjusted based on actual versus expected births.
The distribution of births into male and female for each race group was also assumed to be constant from 2015 through the year 2036. The fraction of the projected births for each age group which was male (or female) was assumed to be the average of the corresponding fractions for the calendar years 2005 through 2015.
- Death Assumptions
These projections use a 2010 unabridged life table for North Carolina developed by this office.
Survival rates are adjusted for actual deaths from 2010-2015.
- Basic Trend Projections by County
The basic technique used to develop county age, race, and sex projections for this series is an Adjusted Migration technique. First, county/state survival rate factors for each county for each race and sex are developed using actual county deaths by race and sex from 2010-2015. These factors are assumed to be reasonably valid throughout the projection period. Then, one obtains a "Cohort Survived" population (assumes no net migration) for each year by applying these factors and state survival rates to the population at the beginning of the year. Finally, one subtracts this "Cohort Survived" population from the population at the end of the year to get the desired survived net migration.
The next step is to adjust this survived net migration for age. The survived net migration for each county for each year for each sex, race and age category is determined by migration trends from the 2000-2010 decade.
Then, the survived net migration for the projection decade is added to this "Cohort Survived" population to yield the final projected population by age.
- Institutional Effects
The basic county trend projections produced for this series were modified for the growth of certain institutions. Institutions such as colleges, universities, military installations, and, to a lesser extent, prisons and some state hospitals, house persons of particular age groups. These populations will substantially grow or decline only by administrative action. There are twelve counties in North Carolina the age structure of which is significantly affected by institutions. These counties (with major institution type) are Avery (prisons and college), Craven (military), Cumberland (military), Durham (university), Jackson (university), Madison (university), New Hanover (university), Onslow (military), Orange (university), Pasquotank (university and prisons), Pitt (university), and Watauga (university).
Many counties have experienced some growth (and in some cases decline) in institutional populations between 2010 and 2015. It was assumed that all institutional populations would remain constant after 2015.
- Projection Controls
Three types of Projection controls were used. The first set was based totally on county estimates. Projected values for July 2010-July 2014, and July 2015 were controlled to revised county estimates for the corresponding dates.
The second set were county population controls for July 1st of each year from 2016 through 2036. They were calculated using exponential smoothing/ARIMA models for the non-group quarters population.
The third set of controls were on race/sex totals for each county. Race and sex totals were projected for each county based on historic growth trends (2000-2010) with the total controlled to the projected county total population.
Last Update: October 14, 2016