Evaluation of simple methods for regional mortality forecasts

Genus

Journal of Population Sciences

Table 2 Input data requirements of the eight regional mortality forecasting methods

Method	National input data	Regional input data
National Death Rates	National forecast death rates	None (Data cells = 0)
SMR Scaling	National base period and forecast death rates	Regional base period SMRs: total deaths by region; ERPs⁽¹⁾ by age and sex by region (Data cells = 41r) ⁽²⁾
Broad Age SMR Scaling	National base period and forecast death rates	Regional base period SMRs: deaths by sex and broad age group by region; ERPs by sex and age group by region (Data cells = 46r)
Mortality Surface	National mortality surface of past and forecast _nL_x values	Projected e₀ by sex by region.⁽³⁾ (Data cells = 2r)
Brass Relational	National base period and forecast l_x values	Regional base period l_x values: deaths and ERPs by age and sex to create life tables. (Data cells = 80r)
Rate Ratio Scaling	National base period and forecast death rates	Regional base period rate ratios: deaths and ERPs by sex and age group by region (Data cells = 80r)
Broad Age Rate Ratio Scaling	National base period and forecast death rates	Regional base period broad age rate ratios: deaths and ERPs by sex and broad age group by region (Data cells = 12r)
TOPALS	National base period and forecast death rates	Regional base period rate ratios: deaths and ERPs by sex and age group by region. (Data cells = 80r)

Notes: (1) ERP = Estimated Resident Population. (2) r = number of regions. (3) If base period regional e₀ values are not available they will need to be calculated, requiring deaths and ERPs by age and sex to create life tables. Regional e₀ forecasts can be created by assuming they remain a constant proportion of an independent national e₀ forecast throughout the forecast horizon