Monthly Japanese trade data December 1998 to December 2008 (extracted from PDFs)

Japan's official monthly trade data are available in machine-readable format from January 2009, but data are available in PDF documents for the period December 1998 to December 2008. For each specific month and flow (export/import), 22 PDF documents are provided, and each of these are multiple pages. The data provided in the PDF files includes monthly trade data in both quantity (up to two units per commodity) and value for all HS codes at 9-digit level. With 121 months, two flows, and 22 files each, this comes to a total of 5324 PDF files.

I have passed these files individually to the Python library Camelot, cleaned the output automatically, and combined these into one CSV file for each month and flow. The raw Camelot output files are included in this archive under 'camelot' subfolders under each month folder. The final CSV files have the filename format YYYYMM.csv.

A very small number of records have been dropped because of poor output from Camelot. These appear to be limited to pages that had a single record and that record was at 2-digit level, meaning that no full-detail data appear to have been missed.

The PDF files were downloaded from e-Stat, the "Portal Site of Official Statistics of Japan":

Exports: https://www.e-stat.go.jp/en/stat-search/files?page=1&layout=datalist&toukei=00350300&tstat=000001013141&cycle=1&tclass1=000001013183&tclass2=000001013184&tclass3val=0&metadata=1&data=1

Imports: https://www.e-stat.go.jp/en/stat-search/files?page=1&layout=datalist&toukei=00350300&tstat=000001013141&cycle=1&tclass1=000001013183&tclass2=000001013185&tclass3val=0&metadata=1&data=1