Estimate Lucas Wedge for Northern Ireland

estimate_lucas_wedge.R

library(purrr)
library(readr)
library(ggplot2)
library(forecast)
library(magrittr)


URI_BASE_PATH <- "https://gist.githubusercontent.com/O1sims/07458f128809996046a5515cad3bf1d1/raw/f92c95d47c9c2d8238c6fb8d41b851eaaff5c1fc"


percentage_change <- function(vector) {
  sapply(
    X = 1:length(vector), 
    FUN = function(x) { 
      if (x == 1) { 
        return(0) 
      } else {
        return((
          (vector[x] - vector[x-1]) / vector[x-1]) * 100)
      }
    }) %>%
    return()
}


calculate_regional_gva <- function(region, gvaData) {
  if (region != 'all') {
    gvaData %<>% subset(Region == region)
  }
  totalRegionalGva <- sapply(
    X = gvaData[, 6:24], 
    FUN = sum) %>% 
    as.vector()
  GvaPercentageChange <- totalRegionalGva %>% 
    percentage_change()
  data.frame(
    year = regionalGvaData[, 6:24] %>% 
      names(),
    totalGVA = totalRegionalGva,
    percentageChange = GvaPercentageChange,
    stringsAsFactors = FALSE) %>%
    return()
}


calculate_uk_gdp <- function(gdp.data, from = 1960) {
  years <- gdp.data$Year %>% 
    unique() %>%
    subset(. >= from)
  yearlyGDP <- sapply(years, function(x) {
    yearlyData <- gdp.data %>% 
      subset(Year == x) 
    sum(yearlyData$GDP) %>% 
      return()
  })
  percentageChange <- yearlyGDP %>% 
    percentage_change()
  data.frame(
    year = years,
    totalGDP = yearlyGDP,
    percentageChange = percentageChange,
    stringsAsFactors = FALSE) %>%
    return()
}


estimate_lucas_wedge <- function(regional.gva.data, uk.gdp.data) {
  lucasWedge <- c()
  projected.gva.data <- regional.gva.data[
    regional.gva.data %>% 
      nrow(), 25:29] %>%
    purrr::flatten_dbl()
  regional.gva <- calculate_regional_gva(
    region = "Northern Ireland", 
    gvaData = regional.gva.data)
  counterfactual.projection <- regional.gva$totalGVA %>%
    forecast::auto.arima() %>%
    forecast::forecast(
      projected.gva.data %>% 
        length()) %>%
    as.data.frame()
  for (i in 1:(projected.gva.data %>% length())) {
    lucasWedge %<>% 
      append(counterfactual.projection$`Point Forecast`[i] - projected.gva.data[i])
  }
  data.frame(
    year = 2017:2021, 
    lucasWedge = lucasWedge) %>%
  return()
}


regional.gva.data <- URI_BASE_PATH %>% 
  paste0("/gross-value-added.csv") %>% 
  read_csv()

uk.gdp.data <- URI_BASE_PATH %>%
  paste0("/gross-domestic-product-data.csv") %>% 
  read_csv()

lucasWedge <- estimate_lucas_wedge(
  regional.gva.data = regional.gva.data, 
  uk.gdp.data = uk.gdp.data)

lucasWedge %>%
  ggplot() +
  geom_line(mapping = aes(x = year, y = lucasWedge)) +
  geom_point(mapping = aes(x = year, y = lucasWedge)) +
  theme_minimal() +
  xlab("") +
  ylab("Deadweight loss (£m)")

gross-domestic-product-data.csv

gross-value-added.csv