<p>Following the concept of utility, this paper examines a way to reduce further the time and cost associated with an already cost-effective approach to population forecasting, the Hamilton–Perry method (H–P method), which not only has been found to produce reasonably accurate forecasts but has a wide range of applications, including stable population theory and historical demography. The usual application of the H–P method, H–P-Usual, is based on gender-specific CCRs and CWRs, using the female population in the childbearing years to compute the CWRs. H–P-Usual has potential drawbacks. If population forecasts by age are only needed for the total population, additional time and resources are required to assemble and evaluate the gender-specific CCRs and CWRs. More pressing, especially for subcounty areas, is the issue of small population sizes in the age and gender-specific population cells. One potential solution to these issues is reducing the number of input cells required by the H–P model. We propose a simpler H–P method, H–P-Light, that uses non-gender-specific CCRs and CWRs. We analyzed 10-year forecast errors for US counties and census tracts and found no degradation of performance using H–P-Light in both counties and census tracts. We conclude that H–P-Light is a viable alternative to H–P-Usual for producing age-specific population forecasts for the total population and, if needed, by gender.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A Simplified Version of the Hamilton–Perry Method for Forecasting Population by Age Group and Gender

  • Jeff Tayman,
  • David A. Swanson

摘要

Following the concept of utility, this paper examines a way to reduce further the time and cost associated with an already cost-effective approach to population forecasting, the Hamilton–Perry method (H–P method), which not only has been found to produce reasonably accurate forecasts but has a wide range of applications, including stable population theory and historical demography. The usual application of the H–P method, H–P-Usual, is based on gender-specific CCRs and CWRs, using the female population in the childbearing years to compute the CWRs. H–P-Usual has potential drawbacks. If population forecasts by age are only needed for the total population, additional time and resources are required to assemble and evaluate the gender-specific CCRs and CWRs. More pressing, especially for subcounty areas, is the issue of small population sizes in the age and gender-specific population cells. One potential solution to these issues is reducing the number of input cells required by the H–P model. We propose a simpler H–P method, H–P-Light, that uses non-gender-specific CCRs and CWRs. We analyzed 10-year forecast errors for US counties and census tracts and found no degradation of performance using H–P-Light in both counties and census tracts. We conclude that H–P-Light is a viable alternative to H–P-Usual for producing age-specific population forecasts for the total population and, if needed, by gender.