Why XBRL needed new formats
Traditional XBRL is XML, which is verbose. That is fine for a few hundred facts, but supervisory datasets now run to hundreds of thousands of rows. Files get large, validation gets slow and the data teams preparing them already work in tables. The Open Information Model separates what the data means from how it is written down, so the same report can be expressed in whichever format suits it (XBRL International).
How xBRL-CSV works
An xBRL-CSV report is a set of CSV files plus a small JSON metadata file. The metadata says what each table and column means: which taxonomy concept, which dimensions, which units. The CSV files hold the rows. Because the mapping is defined by the standard, an xBRL-CSV report converts to XBRL XML and back without losing anything.
xBRL-JSON represents each fact as a JSON object, which suits APIs and developers who want to work with individual facts rather than whole tables.
What it means for supervisors
xBRL-CSV makes it practical to collect much larger datasets under the same taxonomy and rules, which is a step towards granular collection without abandoning XBRL. It is also closer to how firms already hold their data, which lowers preparation cost. The practical questions are whether your validation engine reads these formats natively and whether you can accept both XML and CSV while filers transition.
Where AI fits, and where people decide
The hardest part for a filer is mapping internal tables and columns to taxonomy concepts and dimensions. AI can propose that mapping and explain each choice. A person approves it once, and from then on the mapping runs the same way every time.
- 1Do any of your returns run to tens of thousands of rows or more?
- 2Do filers already produce the data from tables or databases?
- 3Does your validation engine read the Open Information Model formats natively, or convert first?
- 4Can you accept both XML and CSV during a transition period?
- 5Will analysts query the data as tables once it lands?
Where Line4 fits
Line4 is a SupTech platform for financial regulators. It collects returns in XBRL, Excel and CSV, validates them against your taxonomies and rules at the point of submission, and puts AI to work reading them, with supervisors making every decision. If you work with a standard we have not yet run in production, we will say so plainly and show you how we would approach it.
Questions
What is xBRL-CSV?
xBRL-CSV is an XBRL International specification for expressing XBRL data as CSV tables described by a JSON metadata file. It became a Recommendation in October 2021 and suits large, highly structured datasets.
What is the XBRL Open Information Model?
The Open Information Model (OIM) is a syntax-independent model of XBRL data. It allows the same report to be expressed as XML, JSON or CSV and transformed between them without losing meaning.
Is xBRL-CSV still XBRL?
Yes. It uses the same taxonomies and validation as XML-based XBRL. Only the way the data is written down changes.
When should a regulator use xBRL-CSV instead of XBRL XML?
When returns are large or highly structured, such as granular or high-row-count datasets, and when filers already hold the data in tables.