Transforms posterior draws of historical decompositions into a table containing posterior means, medians, standard deviations, and equal-tailed credible intervals, or retains the individual shock contributions when requested.
Value
A bsvar_post_tbl (tibble subclass) with columns model,
object_type, variable, shock, time,
mean, median, sd, lower, and
upper. When draws = TRUE, columns draw and
value replace the summary statistics.
Examples
# Small posterior (S = 5 draws)
data(us_fiscal_lsuw, package = "bsvars")
spec <- bsvars::specify_bsvar$new(us_fiscal_lsuw, p = 1)
#> The identification is set to the default option of lower-triangular structural matrix.
post <- bsvars::estimate(spec, S = 5, show_progress = FALSE)
# Posterior summaries of historical decompositions
result <- tidy_hd(post)
head(result)
#> # A tibble: 6 × 20
#> model object_type variable shock time mean median sd lower upper
#> <chr> <chr> <chr> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 model1 hd ttr ttr 1948.25 -2.38 -2.85 0.764 -3.00 -1.51
#> 2 model1 hd ttr ttr 1948.5 -3.89 -4.63 1.09 -4.72 -2.61
#> 3 model1 hd ttr ttr 1948.75 -4.59 -5.07 1.03 -5.19 -3.17
#> 4 model1 hd ttr ttr 1949 -5.96 -5.70 2.40 -8.92 -3.49
#> 5 model1 hd ttr ttr 1949.25 -9.25 -6.51 8.18 -20.2 -3.77
#> 6 model1 hd ttr ttr 1949.5 -18.0 -7.73 25.4 -52.2 -4.30
#> # ℹ 10 more variables: analysis_type <chr>, interval_probability <dbl>,
#> # interval_method <chr>, scale_by <chr>, scale_var <chr>, scale_factor <dbl>,
#> # units <chr>, source_package <chr>, source_version <chr>, n_draws <int>