The comparative-development toolkit

Inequality, concept by concept

Milanovic separates inequality between countries as units (concept 1) from inequality between countries weighted by their populations (concept 2). inequality() gives every standard measure under either:

inequality(snap$gdp_per_capita)
#> # A tibble: 7 × 2
#>   measure   value
#>   <chr>     <dbl>
#> 1 gini      0.638
#> 2 theil_t   0.746
#> 3 theil_l   0.916
#> 4 atkinson  0.600
#> 5 cv        1.54 
#> 6 palma     9.36 
#> 7 p90_p10  43.7
inequality(snap$gdp_per_capita, weights = snap$population)
#> # A tibble: 7 × 2
#>   measure   value
#>   <chr>     <dbl>
#> 1 gini      0.612
#> 2 theil_t   0.686
#> 3 theil_l   0.774
#> 4 atkinson  0.539
#> 5 cv        1.40 
#> 6 palma     7.70 
#> 7 p90_p10  32.7

How much of it lies between continents? Theil’s L decomposes without depending on the order:

theil(snap$gdp_per_capita, snap$population, groups = snap$continent, type = "L")
#> # A tibble: 3 × 3
#>   component value share
#>   <chr>     <dbl> <dbl>
#> 1 total     0.774 1    
#> 2 between   0.349 0.451
#> 3 within    0.425 0.549

Convergence and mobility

beta_convergence() and sigma_convergence() summarise a distribution in one number each. transition_matrix() shows how countries move within it (Quah):

set.seed(4)
pan <- expand.grid(iso3c = snap$iso3c[1:60], year = 2000:2010,
                   stringsAsFactors = FALSE)
pan$gdp <- exp(9 + stats::ave(stats::rnorm(nrow(pan), 0, 0.05), pan$iso3c,
                              FUN = cumsum) + rep(stats::rnorm(60), 11))
tm <- transition_matrix(pan, gdp, n_classes = 3, classes = "relative")
tm
#> # A tibble: 9 × 4
#>    from    to     n      p
#>   <int> <int> <int>  <dbl>
#> 1     1     1   189 0.940 
#> 2     2     1    12 0.0594
#> 3     3     1     0 0     
#> 4     1     2    12 0.0597
#> 5     2     2   180 0.891 
#> 6     3     2     5 0.0254
#> 7     1     3     0 0     
#> 8     2     3    10 0.0495
#> 9     3     3   192 0.975
attr(tm, "ergodic")
#> [1] 0.2521957 0.2534504 0.4943538
rank_mobility(pan, gdp, from = 2000, to = 2010)
#> # A tibble: 1 × 4
#>       n   tau  p_value share_moved
#>   <int> <dbl>    <dbl>       <dbl>
#> 1    60 0.879 3.27e-23       0.217

Money across countries and years

deflate() turns current prices into constant ones, and to_ppp() converts exchange-rate dollars to purchasing-power parity; a cross-country panel over time usually needs both. Both fetch their price series from the World Bank:

d <- country_data(2010:2020, c(gdp = "NY.GDP.MKTP.CN"), panel = TRUE)
d |> deflate(gdp, base_year = 2015) |> to_ppp(gdp_real)