benchmark {povmap}

Benchmark Function

Description

This function benchmarks the EBLUP estimates of an area-level model.

Usage

benchmark(object, benchmark, share, type = "raking", overwrite = FALSE)

Arguments

object

an object of type "fh".

benchmark

a number determining the benchmark value.

share

a vector containing the shares of the population size per area and the total population size (N_d/N).Values must be sorted like the domains in the fh object.

type

Character indicating the type of benchmarking. Types that can be chosen (i) Raking ("raking"), (ii) Ratio adjustment ("ratio"), (iii) MSE adjustment ("MSE_adj"). Defaults to "raking".

overwrite

if TRUE, the benchmarked FH estimates are added to the ind object of the emdi object and the MSE estimates are set to NULL since these are not benchmarked. Defaults to FALSE.

Details

The benchmarking algorithm only works, if FH estimates are available. The type "MSE_adj" only works, if MSE estimates are available. If overwrite is set to TRUE, the emdi object is returned, but the benchmarked FH estimates are added to the ind object of the emdi object and the MSE estimates are set to NULL since these are not benchmarked.

Value

A data frame containing a domain indicator (Domain), direct estimates (Direct), point predictions (FH), benchmarked point predictions (FH_Bench) and a variable indicating out-of-sample domains Out (1 for out-of-sample, 0 for in-sample) . If overwrite is set to TRUE, the fh object is returned, but the point predictions of the ind data frame are complemented by the benchmarked results.

References

Datta,G. S., Ghosh, M., Steorts, R. and Maples, J. (2010) Bayesian benchmarking with applications to small area estimation. Test, 20, 574–588.

Examples

# Loading data - population and sample data
data("eusilcA_popAgg")
data("eusilcA_smpAgg")

# Combine sample and population data
combined_data <- combine_data(
  pop_data = eusilcA_popAgg,
  pop_domains = "Domain",
  smp_data = eusilcA_smpAgg,
  smp_domains = "Domain"
)

# Estimate Fay-Herriot model
fh_std <- fh(
  fixed = Mean ~ cash + self_empl, vardir = "Var_Mean",
  combined_data = combined_data, domains = "Domain", method = "ml",
  MSE = TRUE
)

# Benchmark the point estimates

# Example 1: Receive data frame with point estimates and their benchmarked
# results
fh_bench <- benchmark(fh_std,
  benchmark = 20140.09,
  share = eusilcA_popAgg$ratio_n, type = "ratio"
)

# Example 2: Add benchmarked results to fh object
fh_bench <- benchmark(fh_std,
  benchmark = 20140.09,
  share = eusilcA_popAgg$ratio_n, type = "ratio", overwrite = TRUE
)