Birth rate forecast based on the FSO methodology • propopbirth

Birth rate forecasts based on FSO (Federal Statistical Office) methodology.

Overview

The propopbirth package consists of several modules: First, model input data are calculated. Second, TFR (total fertility rate) and MAB (mean age of the mother at birth) are predicted. Third, TFR and MAB forecasts are used to predict the age-specific birth rates.

Population projections

The output from propopbirth is fully compatible with {propop} - An R package for population projections in R. {propop} needs future birth rates as input. This is only available at the FSO at cantonal level. With propopbirth the birth rates can also be calculated for other spatial units or with own hypothesis.

Installation

To install the current github version of the package, make sure you have devtools installed and type:

devtools::install_github("statistik-aargau/propopbirth")

Vignettes

The package includes three vignettes.

Vignette 1: Prepare model input data from birth and population, with examples
Vignette 2: Forecast of TFR (total fertility rate) and MAB (mean age of the mother at birth), with examples
Vignette 3: Forecast of the age-specific fertility rate; including an example of the complete forecast of the birth rate

Limitations, future plans

Limitations

With population scenario models, it is always necessary to check whether the spatial units are large enough to ensure that meaningful forecast estimates. There is no population limit specification by the FSO, when not to run the forecast models. However, it visual checks are advisable (graphs see vignettes for examples). These are helpful to verify whether the estimates for the birth rate forecasts are meaningful.

Future plans

Currently, FSO birth data is not accessible with the required variables. For this reason, the birth data is included in the propopbirth package. As soon as the birth data is publicly accessible (similar to population data), it will be directly accessible in propopbirth . This will allow a more flexible selection of spatial units (not only municipalities, but also districts).

Examples

Detailed examples with explanations are provided in the vignettes (step by step). Here an example of entire birth rate forecast for three municipalities.

Birth and population data

library(ggplot2)
library(propopbirth)
library(dplyr)

Create model input data

input <- create_input_data(
  population = fso_pop,
  births = fso_birth |> 
    dplyr::filter(spatial_unit %in% c("Aarau", "Frauenfeld", "Stadt Zürich")),
  year_first = 2011,
  year_last = 2023,
  age_fert_min = 15,
  age_fert_max = 49,
  fert_hist_years = 3,
  binational = TRUE
)

TFR (total fertility rate) forecast

forecast_tfr <- forecast_tfr_mab(
  topic = "tfr",
  topic_data = input$tfr,
  trend_model = c(
    model = "lm", start = 2024, end = 2026, trend_past = 7, trend_prop = 0.5
  ),
  temporal_model = c(
    model = "cubic", start = 2027, end = 2055, trend_prop = 0.8, z0_prop = 0.7,
    z1_prop = 0
  ),
  temporal_end = NA,
  constant_model = c(model = "constant", start = 2056, end = 2075)
)

MAB (mean age of the mother at birth) forecast

forecast_mab <- forecast_tfr_mab(
  topic = "mab",
  topic_data = input$mab,
  trend_model = c(
    model = "lm", start = 2024, end = 2026, trend_past = 7, trend_prop = 0.5
  ),
  temporal_model = c(
    model = "Bezier", start = 2027, end = 2055, trend_prop = 0.3, z0_prop = 0.7,
    z1_prop = 0
  ),
  temporal_end = NA,
  constant_model = c(model = "constant", start = 2056, end = 2075)
)

Forecast of the age-specific fertility rate

forecast_fer <- forecast_fertility_rate(
  fer_dat = input$fer,
  tfr_dat = forecast_tfr,
  mab_dat = forecast_mab, 
  year_begin = 2024, 
  year_end = 2075)

Acknowledgment

We acknowledge the Federal Statistical Office (FSO) for providing their SAS code used for fertility projections. And we are grateful to the FSO for giving the permission to use the birth data within the propopbirth package.