real_wages_decade.R

library(auseconhist) # remotes::install_github("MattCowgill/auseconhist")
library(readabs)
library(tidyverse)
library(slider)

butlin <- auseconhist::butlin_t7 |> 
  filter(!is.na(awe),
         year >= 1901) |> 
  select(date = year, value = awe, cpi) |> 
  mutate(series = "awe",
         source = "Butlin, Dixon & Lloyd (2014)",
         date = ymd(paste(date, "06", "01", sep = "-")))

awe <- read_awe("awe")
awote <- read_awe("awote")
cpi <- read_cpi()

abs <- awe |> 
  bind_rows(awote) |> 
  group_by(series = wage_measure) |> 
  padr::pad("day") |> 
  fill(value) |> 
  select(date, value, series) |>
  mutate(source = "ABS") |> 
  left_join(cpi, 
            by = "date") |> 
  filter(!is.na(cpi)) 

butlin |> 
  bind_rows(abs) |> 
  group_by(series, source) |> 
  mutate(real_awe = (value / cpi) * cpi[date == ymd("2010-06-01")]) |> 
  mutate(real_awe_10yago = slide_index_dbl(.x = real_awe,
                                           .i = date,
                                           .f = \(x) x[1],
                                           .before = years(10),
                                           .complete = TRUE)) |> 
  mutate(real_awe_growth = (real_awe / real_awe_10yago) - 1) |> 
  arrange(real_awe_growth)

Sure thing @dylanjmcconnell

There are 3 periods in which real wages have fallen: the Great Depression (decade to the mid-30s), the 90s recession (mid-80s to mid-90s) and 2013-2023.

The fact that the past 10 years has anything in common with those two other periods is damning in itself, and makes the point they want to make. But - unless they're using some other data source I'm aware of - the claim that this is the first 10 year period in which real wages have fallen is not correct.