Leveraging Inline XBRL for Insurance Pillar III Disclosures: a Proof of Concept

Posted on May 5, 2021 by Editor

This is a guest post by Antoine Bourdais and Clément Duhamel of Invoke. It is based on their fantastic live demonstration at Data Amplified Virtual on 16 April 2021, showing how we can use current technology, facilitated by recent regulatory advances, to collect and analyse public Pillar III information disclosed by the European financial industry, making it truly digital, structured and useful.

Public information can and should be digitised to facilitate access, analysis and comparison. While reporting around the world is digitising fast, there is still important data that is not being published digitally today and is therefore not available in useful ways. We are interested in whether we can use existing tools to digitise this information and take a look into its future. Our proof of concept discussed here considers the application of state-of-the art analytical technology to banking and specifically insurance Pillar III disclosures, leveraging recent advances in regulatory requirements and the introduction of Inline XBRL (or iXBRL) reporting in Europe.

Europe and XBRL

The three European Supervisory Authorities (ESAs) – the European Banking Authority (EBA), the European Insurance and Occupational Pensions Authority (EIOPA), and most recently the European Securities and Markets Authority (ESMA) – all require XBRL in regulatory reporting. The European banking and insurance industries have therefore been very familiar with using XBRL to share data with regulators (EBA and EIOPA) for a number of years now. However, this risk and compliance information is not available to the public.

At the same time, these sectors do require certain public Pillar III disclosures, under the Capital Requirements Directive (CRD V) in the case of the EBA and Solvency II for EIOPA. There is no mandatory format for this information. CRD V has only recently come into effect so there is as yet no public banking data available, but Solvency II reports are generally published as PDF documents. While a PDF is of course a digital rather than a physical document, it’s really the electronic version of paper – it does not contain digital, machine readable data.

Nonetheless, both sets of regulations do provide reporting templates as guidelines for the publication of Pillar III disclosures. In the insurance sector, the Solvency II taxonomy already includes these guidelines, which makes the information easier to digitally extract and compare.

At the same time, this year has seen the introduction of the European Single Electronic Format (ESEF) for financial reporting by issuers on EU regulated markets. This mandates the use of Inline XBRL in public annual reports, although these are voluntary in many countries this year due to the Covid-19 crisis.

So why explore this particular proof of concept, using Inline XBRL for insurance Pillar III disclosures? First, because ESEF is now a digital reality, generating the tools and experience needed to incorporate Inline XBRL data in a single report package that is readable by both machines and humans. On the other hand, Pillar III public disclosures offer us a domain to explore that is not yet digital. To help us do that we have the PDFs produced by insurance companies as our raw material, and the corresponding EIOPA Solvency II taxonomy.

We will show that implementing Inline XBRL for narrative report production and analysis of public information is not a difficult issue from a technological point of view, and brings clear benefits.

Report production and validation

We leveraged our existing Invoke ESEF Preparation Tool, using the Solvency II taxonomy to handle insurance data. The integration of the EIOPA templates directly from the taxonomy allows us to tell the computer what to look for, and what the data means.

Once this model is in place, we can very easily import data from individual Pillar III reports, known in the case of the insurance sector as Solvency and Financial Conditions Reports, or SFCRs. All we need to do to create a truly digital report is create a new project and load a PDF, which immediately converts directly into an XHTML document. To automatically add the XBRL layer, we then go through the report selecting the relevant tables as project sources. The existence of reporting template is very important in facilitating the process here; for each selected table in the PDF we simply identify which template table it corresponds to, and therefore what data it should contain.

As we map tables from the PDF to the regulatory disclosure requirements captured in the template, the system carries out validation processes, checking data quality and highlighting inconsistencies.

After reviewing any detected issues, we can automatically generate an Inline XBRL report package. To a human, this looks exactly like the original PDF, but it is a genuinely digital document. Using a viewer like the Invoke XBRL Viewer the reader is able to navigate into the XBRL layer and dig into data, getting a better idea of the meaning behind the numbers. The report is also machine readable, allowing the information to be captured, analysed and compared.


Having captured the publicly available Pillar III reports published by insurance companies in PDF, and generated digital, validated information from them, perhaps the more innovative aspect of our proof of concept is the analysis of the data.

We used the process described above to create XBRL reports from all the SFCR disclosures, available for about 20 companies. We then simply prepared our XBRL analytical platform to collate, render and analyse all the data together, into which we loaded the reports.

To illustrate this proof of concept, we leveraged existing analysis disclosed on the EIOPA website, which is based on non-public regulatory data. This allows us to compare the different insurance companies, to look at the distribution of data points over the sector as a whole, and to slice the data up to analyse it in as many different ways as we can think of. For example we can compare assets between companies, view liabilities and technical provisions for all our entities, or focus in on solvency capital requirements (SCR).

The existence of the XBRL taxonomy is crucial to this kind of comparative analysis, providing a common ‘dictionary’ to all the reporting companies. But since they are currently submitting that information to the public in PDF, it is not easy to benefit from what is available. Digitisation makes it straightforward for analysts to get the most out of the data, using existing tools.

The technology is ready

We certainly came up against some issues during the digitisation process. It became clear to us that some companies today are not necessarily publishing the required level of public information. That is not easy to detect in PDF format, but is very obvious when we try to generate digital data. Errors also emerge. For example, some companies have employed the wrong template, using the solo template for group information. We also see frequent errors relating to currencies and accuracy. Digitisation therefore helps to evaluate and benchmark data quality between different issuers.

It is also very clear, however, that the technology we have available is mature enough for the digitisation of public information, both in terms of production and of analysis. We hope that we’ve demonstrated the value of digitisation in terms of greater transparency and utility of data, and at the same time shown that the transition to Inline XBRL and digital information is really not a huge challenge for companies. The large amount of digital information that is already being prepared for regulators is to a very great extent reusable for Pillar III disclosures, and would make them significantly more useful.

Finally, we would like to highlight the fact that Inline XBRL is a very smart vehicle to support this digitisation of narrative reports, with effective tools and excellent results.

We’re bringing you a range of insights and reflections from our world-leading speakers – check out the Data Amplified tag to see more, and watch this space!

Other Posts


Would you like
to learn more?

Join our Newsletter mailing list to
stay plugged in to the latest
information about XBRL around the world.

By clicking submit you agree to the XBRL International privacy policy which can be found at xbrl.org/privacy