What is validation?
Validation is the process of carrying out automated data quality checks on XBRL reports – ‘validating’ the data against testable rules that it should follow, and enabling potential errors to be detected and fixed.
Validation is an intrinsic part of XBRL, and is essential to ensuring high data quality and consistency in XBRL reports. Digital reporting creates information that is computer-readable, comparable and decision-useful, retaining its integrity throughout the reporting chain, but mistakes can be introduced during report preparation. Validation rules are designed to detect and prevent tagging errors and inconsistencies in the data across a report.
A layered approach
XBRL validation comes in several layers.
- Fundamental levels of validation are built into the core XBRL specification. XBRL software must ensure that a reported fact matches its data type and has the required attributes – for example, a monetary amount must be a number, and must come with an appropriate currency unit.
- The XBRL Calculations specification can add a second layer of rules dealing with simple calculation relationships between facts, such as “assets = current assets + non-current assets.” If the figures do not add up, software can flag a possible issue.
- More in-depth validation is done using additional validation rules, sometimes called business rules, typically via the XBRL Formula specification. These diverse rules can capture a wide range of conditions that reported data should meet. Each set of validation rules is different, tailored to the reporting rules they serve and the type of information needed.
- Advanced reporting systems may add a fourth level in their approach to validation, whereby they continually monitor data quality outcomes, identifying common issues and why they occur. Quality is built into the data collection framework, enabling ongoing improvement.
How does validation work?
The information needed for validation is included in XBRL taxonomies – the digital dictionaries used for reporting. Most taxonomies incorporate specific validation rules in addition to the basic rules built into XBRL.
Before an XBRL report is published, or submitted to a regulator, the computer-readable tagged data should be checked by software against the validation rules. Validation capabilities are included in XBRL software, and most regulators carry out their own validation checks against business rules.
When a validation issue is flagged by software, it doesn’t always indicate that something is wrong, but it should prompt report issuers to examine the data. A validation check has three possible results, with different ‘severities:’
- OK – when the data meets validation requirements and no problem is detected.
- Warning – indicating a minor issue that should be reviewed, but will not prevent report submission.
- Error – for a serious data-quality concern which requires correction, and will normally cause a report to be rejected.
Ideally, companies should validate their draft disclosures on an ongoing basis throughout the preparation process, enabling them to detect and correct problems early on.
Validation rules can take a wide range of forms, depending on the needs of the data. Just a few examples of possible rules include:
- Many facts need to add up to a total, requiring simple relationship-based rules of the “a+b=c” type.
- A number may be required to have a positive (or negative) value – preventing sign errors, one of the most frequent reporting issues.
- Dates may need to fall within a certain range, or to be consistent with other dates in the same report.
- A percentage may, in some (but by no means all) cases, be required to have a value from 0-100.
- Some rules are designed to check reports for missing facts, where these have been accidentally or even deliberately omitted.
- When standards are updated, validation rules can highlight the use of outdated tags.
It is worth noting that validation rules do not check the factual accuracy of data, simply whether it meets specific conditions. However, they do tend to catch many common errors.
Where do validation rules come from, and what does Formula do?
XBRL Formula, Calculations and the core built-in XBRL validation are all part of the XBRL standard, which is developed and maintained by XBRL International and its global network of expert volunteers.
The XBRL Formula specification provides a standardised mechanism for defining validation rules within an XBRL taxonomy, using a standard format that can be understood and implemented by any XBRL software.
Formula rules range from very simple to highly sophisticated. Formula can be used for all of the examples above, but it can also be used for much more intricate rules reflecting the complexities of reported information and the relationships between multiple concepts. For example, the value of one fact may determine whether other facts are needed, such as when larger companies must report more information. A rule could check that companies whose revenue or number of employees exceeds a given threshold have reported and tagged a range of required data, while not requiring the same facts of smaller companies.
Taxonomy authors use XBRL Formula to build their own set of validation rules, specific to their own reporting framework and the data they need to collect. Most taxonomies are created by standards setters, regulators and similar national and international bodies, such as central banks and tax authorities. For guidance on how to create validation rules, you can read our XBRL Formula Rules Tutorial here.
Read more
- More on XBRL taxonomies
- Taxonomy authors can find guidance on creating validation rules in our XBRL Formula Rules Tutorial
- XBRL US plays an active role in promoting data quality: filers to the US SEC can materially improve the quality of their filings using validation rules developed by the XBRL US Data Quality Committee (DQC)
- One example of continuing monitoring and feedback of validation data to improve the reporting process and build data quality is found in the European Central Bank’s supervisory reporting.
- Some common tagging errors, found in our analysis of ESEF filings, are explored in our blog series here
- Read our latest news items on validation here
