ffiec.pq

ffiec.pq provides tools for processing FFIEC Call Report bulk data into Parquet files that can be analyzed from R, Python, DuckDB, and other columnar data tools.

The package is designed around a local data repository:

• raw FFIEC bulk zip files live under RAW_DATA_DIR/ffiec
• processed Parquet files are written under DATA_DIR/ffiec

If RAW_DATA_DIR or DATA_DIR is not set, ffiec.pq will ask for the directory the first time it needs it in an interactive R session.

Installation

install.packages("pak")
pak::pak("iangow/ffiec.pq")

Basic workflow

Process the FFIEC bulk files:

library(ffiec.pq)

# If RAW_DATA_DIR or DATA_DIR is not already set, this will prompt for them.
results <- ffiec_process(use_multicore = FALSE)

Read the resulting Parquet data lazily through DuckDB:

library(DBI)
library(dplyr)
library(duckdb)

db <- dbConnect(duckdb::duckdb())

por <- ffiec_scan_pqs(db, "por")
por |>
  count(date) |>
  arrange(desc(date)) |>
  collect()

See the package website article “Data curation: the case of Call Reports” for a longer discussion of obtaining, processing, and using the data.