add prepare_sector_split.R

helper_functions.R

@@ -0,0 +1,69 @@
+lost_companies_sector_split <- function(abcd,
+                                        companies_sector_split) {
+  abcd_id <- abcd %>%
+    dplyr::distinct(.data$company_id, .data$name_company)
+
+  # identify lost_companies_sector_split and write to csv for inspection
+  lost_companies_sector_split <- companies_sector_split %>%
+    dplyr::anti_join(
+      abcd_id,
+      by = c("company_id")
+    )
+
+  return(lost_companies_sector_split)
+}
+
+apply_sector_split_to_loans <- function(data,
+                                        abcd,
+                                        companies_sector_split) {
+  unique_companies_pre_split <- data %>%
+    distinct(name_abcd)
+
+  abcd_id <- abcd %>%
+    dplyr::distinct(.data$company_id, .data$name_company)
+
+  companies_sector_split <- companies_sector_split %>%
+    dplyr::left_join(
+      abcd_id,
+      by = c("company_id")
+    ) %>%
+    dplyr::select(-"company_id")
+
+  data <- data %>%
+    dplyr::inner_join(
+      companies_sector_split,
+      by = c("name_abcd" = "name_company", "sector_abcd" = "sector")
+    ) %>%
+    dplyr::mutate(
+      # renaming the loan_id is not conditional to avoid any chance of accidentally
+      # renaming a split loan to a loan_id that already exists elsewhere
+      id_loan = paste(.data$id_loan, .data$sector_abcd, sep = "_"),
+      loan_size_outstanding = dplyr::if_else(
+        is.na(.data$sector_split),
+        .data$loan_size_outstanding,
+        .data$loan_size_outstanding * .data$sector_split
+      ),
+      loan_size_credit_limit = dplyr::if_else(
+        is.na(.data$sector_split),
+        .data$loan_size_credit_limit,
+        .data$loan_size_credit_limit * .data$sector_split
+      )
+    ) %>%
+    dplyr::select(-"sector_split")
+
+  unique_companies_post_split <- data %>%
+    distinct(name_abcd)
+
+  if (nrow(unique_companies_pre_split) != nrow(unique_companies_post_split)) {
+    warning(
+      glue::glue(
+        "Applying the sector split has lead to changes in the number of unique
+        companies covered in the analysis. Prior to the split, there were
+        {nrow(unique_companies_pre_split)} unique companies. After the split,
+        there are {nrow(unique_companies_post_split)} unique companies."
+      )
+    )
+  }
+
+  return(data)
+}