THE DELAWARE POPULATION CONSORTIUM
Projections Methodology
DPC Home · Projections · History · Methodology · Notes · Meetings
The projection process begins with the 2000 age-race-gender distribution (obtained from the US Bureau of Census) for the jurisdiction being projected. This distribution is comprised of 101 separate age cohorts beginning with the 0 age group and ending with the 100-and-over age group. The projection process uses the modified age-race-gender (MARS) estimates produced during 2002 to correct certain flaws in the estimates prepared by the Census Bureau in 2000. It also reflects the 2000-2005 age-race-gender distributions provided by the US Centers for Disease Control and Prevention (CDC) for single race categories.
Appropriate survival rates are applied to each single-year age-race-gender cohort. A ten-year survival rate is used to account for mortality. Each age group has its own unique survival rate. Applying the appropriate survival rate to the 0 age group, for example, yields the likely number of those persons alive in that age group in 2000 will survive to enter age 1 in 2001. This procedure is followed until all age groups have been survived with the exception of the birth cohort (age 0 in 2001).
The 0 age group in 2005 is formed from all of those persons expected to be born during the 2004-2005 period. This is estimated by applying an annual birth rate for females in each separate age group beginning with 10 and ending with 49. For example, if the single-year fertility rate for females of age 30 is multiplied by the number of females in that age group in the base year, the result is the likely number of births for those females over the one-year projection period. This is done for all relevant age groups. These estimates are summed to produce the new birth cohort. The birth cohort is then survived to 2001. This procedure is applied for each subsequent time period from 2002 to 2030. The end result of this two-step process is a projection of population growth due solely to natural increase. It must be adjusted for net migration.
The conventional procedure to adjust the projections for net-migration is to calculate a migration rate for each age-race-gender category for some past time period (such as 1995-2000), and to make assumptions about how those rates will change in the future. This procedure works well if the migration rates from the reference period accurately predict the future. If the nature of migration is changing, the result will be unsatisfactory. The rapid change in migration during the 1980's, for example, stands in stark contrast to that of the 1970's. For that reason, and the need to integrate employment projections with the population projections, another method is used to estimate migration (see Migration).
Mortality
Mortality rates are based on age-by-sex-by-race, single-year survival rates calculated from data on the deaths of Delaware residents from 2000-2005 and the underlying population for each cohort in the 2000-2005 interval. Using these rates as a starting point, the rates are scaled proportionally to produce the estimated number of deaths 2000-2005 using actual deaths through 2005. In addition, actual deaths for 2006 and the estimated deaths for 2007 were allowed to influence the 2007 projection. This scaling procedure was followed separately for the State, each of the counties, and the City of Wilmington, resulting in a slightly different set of mortality assumptions for each. Projections for the cities of Newark and Dover utilize the county-wide rates since the populations are too small to produce stable rates.
Fertility
The assumptions for fertility rates were derived using births to Delaware residents from 1996-2005 and the underlying female population groups from the same period. Using these as a base, the rates are scaled proportionally to produce the estimated number of births 2001-2007 using actual births through 2006. Following the same procedures as in mortality above, estimated births in 2007 were allowed to influence the 2007 projection. This scaling procedure was followed separately for the State, each of the counties, and the City of Wilmington, resulting in a slightly different set of fertility assumptions for each. Projections for the cities of Newark and Dover utilize the county-wide rates since the populations are too small to produce stable rates.
Labor-force
Resident employment is calculated to balance the employment forecast with the population projections. This calculation requires a number of assumptions. The first is taken from the Bureau of Labor Statistics (BLS) which provides the current and projected labor-force participation rates for each age-sex cohort from 2000 through 2020. Applying these rates allows the derivation of a theoretical labor-force. An unemployment rate is then applied to that theoretical labor-force. That rate is the same as the one used in generating the employment forecasts. The result of this calculation is employment by place of residence. Resident jobs are derived from employment by place of residence by making adjustments for self-employment, agricultural jobs, and workers with multiple-jobs. BLS also provides labor force estimates from the CPS, CES, and CEW. These blended to update current employment.
Bureau of Labor Statistics estimates and forecasts are used to establish an employment starting point. A single adjustment is required to generate an estimate for resident jobs from the employment forecasts. Net commuting (i.e., in-commuters less out-commuters) is used to adjust the employment forecast.
Migration
Employment (i.e., number of jobs) is usually forecast using an economic model as opposed to a demographic model. It is assumed that there will be a sufficient number of persons to fill those jobs. If the projected population due to natural increase is too low, then net in-migration is required to fill the jobs. Alternatively, if the projected population is too high, net out-migration will occur. In either case, the migration rates from the reference year are modified proportionally to produce sufficient net migration to allow the employment forecasts and the population projection to balance. The final amount of net migration is an output of that calculation. If that estimate seems too high, adjustments must be made to the employment forecasts or to one of the other assumptions in the labor-force estimate.
Adjustments
Projecting population is a risky business and is dependent on the quality of the assumptions. Those can change from year to year as better data becomes available. Thus, it is imperative that the projections be reviewed annually. This report is the result of the Consortium's continuing monitoring and review process.
There are data that can be used to adjust the path of the projection without going through a major revision. First, the annual population estimates (not forecasts) provided by the Federal State Cooperative for Population Estimates (FSCPE) are useful for monitoring the total population. Second, actual birth and death statistics issued by National Center for Health Statistics (NCHS) and the Delaware Health Statistics Center are quite valuable. Third, school enrollment data, estimated migration from the Bureau of Census and the IRS, Medicare enrollment data from HCFA, and the data from the Census Bureau's Current Population Survey (CPS) and American Community Survey (ACS) can help guide structural changes in the age groups. Finally, data from the Department of Labor tracks changes in employment growth. It is imperative that these indicators are tracked and that the projections are modified annually. These projections were carefully benchmarked to known data in 2007 (e.g. actual births, deaths, and estimated migration for the 2000-2006 period).
