Keeping European supervision on Track

Good morning, everyone.

It’s a pleasure to be here in The Hague.

Looking at this international audience, I gather many of you arrived here by train. Maybe before or after taking a plane. Or before or after traveling by car or tram.

Today, trains are far from the first thing that comes to mind when we think about new technologies. But there was a time when trains were cutting-edge.

In The Hague, that time came in 1843, when the very first train arrived.  That’s more than 180 years ago.

The real novelty, however, was not the train itself. It was that The Hague was now connected to Leiden, Haarlem, and Amsterdam in a way that had never been possible before. Perhaps not even imaginable.

Since then, many more connections have been added.

And that is where the real value lies. In the network that gradually emerged around those first train lines.

Every new connection increased the usefulness of the system as a whole. A network that allowed people, goods and ideas to move efficiently between different places.

And this network didn’t stop at borders. Trains could soon connect people from all over the country. And across the borders. Across the continent, really.

When everything runs smoothly, which, let’s be honest, is not always the case, but when it does, it’s easy to forget that one train track can take you from the north of Holland to the south of France. From the British countryside to the hustle and bustle of Brussels.

Now, I’m not here to discuss the means of transportation that got us together today, but to talk about data and how we can share it. Across individuals. Across institutions. Across borders. And how to do this in a simple, smart and structured way.

This is one of the key challenges facing supervisors, financial institutions and policymakers alike. And if we succeed, we will go from data to insight.

There is certainly no shortage of data. Financial institutions generate vast amounts of information. Supervisors increasingly rely on that information in their work. And new technologies create ever more opportunities to collect, process and analyse that data.

At the same time, expectations continue to rise.

We want supervision to become more effective. We want reporting processes to become more efficient. We want insights to become more timely. And increasingly, we expect technology, including artificial intelligence, to help us achieve such goals.

But data does not automatically lead to insight.

In the nineteenth century, Spain's railways were built just a little wider than those in most of Europe. Just a few centimetres. Hardly noticeable. Yet for more than a century, those centimetres meant that trains from Paris, crossing the Spanish border, could not simply carry on to Madrid. Passengers changed trains. Freight changed wagons. A small difference in standards created a very large obstacle. Sometimes the success or failure of integration can depend on details as small as the distance between two rails.

And so it goes for data.

For data to be useful, those who receive it must understand it in the same way as those who provide it.

That requires something more than the mere exchange of data. It requires a shared understanding of meaning. A shared track.

For supervisors, this is a very practical issue.

At DNB, effective supervision depends on our ability to receive, validate, process and analyse large volumes of information. Reporting institutions invest significant resources in producing this information. Both national and European authorities rely on it to fulfil their responsibilities.

As the scale of reporting increases, the challenge is no longer just about moving data from one organisation to another. The real challenge lies in ensuring that the meaning of information remains consistent throughout the entire chain.

That is where standards become important. And that is where XBRL has made a significant contribution.

For many years, XBRL has provided a structured approach to describing information, validation rules and reporting frameworks. It has helped create an environment in which reporting institutions and supervisors can work with the same definitions, the same rules and the same visualisation of reported information.

The benefits are substantial.

Information can be collected, validated and processed in a highly automated way.

Validation rules improve data quality and identify problems early.

And visualisation helps supervisors and reporting institutions quite literally see the same thing, giving them a common basis for discussion.

XBRL supports trust across the reporting chain.

Trust in the sense that reporting institutions, national supervisors and European authorities can work based on the same understanding of the information being exchanged. That understanding is derived solely from the XBRL taxonomy – no external information is needed.

And also trust in the technology used: XBRL International certifies software to work exactly according to specifications. Supervisors and reporting institutions can safely use different software packages, confident of uniform standards.

This may sound straightforward, but anyone who has worked with large reporting systems will know that this is not something that can be taken for granted.

Standards reduce ambiguity and thereby minimise the risk of different parties interpreting information in different ways.

In doing so, they allow organisations to spend more time analysing information and less time interpreting it.

The value of these standards is not limited to a single organisation or a single part of the reporting chain.

The gains accumulate across the entire system. Reporting institutions, supervisors and software vendors all benefit from a common set of specifications. Ultimately, society benefits because resources can be directed towards understanding risks rather than managing complexity.

This systemwide perspective is important. Reporting is not a collection of isolated activities. It is a network.

Changes introduced in one part of that network often affect every other part as well. That is why common standards are so valuable: they provide a shared foundation for cooperation.

This becomes increasingly important as reporting networks become more complex.

Across Europe, supervisors are working with larger datasets, more granular datasets and increasingly interconnected datasets. Initiatives such as the ECB’s Integrated Reporting Framework (IReF) are part of this development.

These developments create significant opportunities.

Greater granularity can improve our understanding of risks. More integrated information can strengthen cooperation between authorities. More powerful analytical tools can help identify patterns that might otherwise remain invisible.

At the same time, growing complexity increases the importance of standardisation.

This becomes especially clear when we consider the role of artificial intelligence.

There is considerable interest in its potential application in reporting, supervision and financial services more broadly.

That interest is justified.

AI has the potential to improve the way information is processed, analysed and used.

However, there is sometimes an implicit assumption that increasingly capable AI systems will compensate for weaknesses in data structures.

In practice, the opposite is often true.

AI tends to perform best when information is well structured, clearly described and accompanied by meaningful context.

As always, the quality of the output depends on the quality of the input.

The more data we collect, the more essential shared understanding becomes.

And it becomes even more essential the more we rely on AI to analyse that data.

This is one reason why the future development of XBRL is so important.

As the standard evolves, it should continue to make information easy to interpret, not only by people, but also by machines.

If we want AI applications to support reporting and supervision effectively, they must be able to make full use of the semantics, metadata and context contained within XBRL taxonomies.

This has practical implications for the future development of reporting standards.

As datasets become larger and more granular, there may be a temptation to focus primarily on the efficient exchange of facts and values.

Efficiency is important, but not at the expense of meaning.

One of the strengths of XBRL is that this meaning is embedded in the reporting process itself.

Artificial intelligence does not reduce the importance of standards. It increases the value of standards that successfully preserve shared understanding.

This is an important consideration for Europe.

European supervision increasingly operates within an interconnected environment involving institutions, national authorities and European authorities. That environment depends on cooperation, interoperability and trust.

Maintaining these qualities requires continued investment in shared infrastructure.

Some forms of infrastructure are highly visible. Like railway infrastructure.

Others are not. Reporting standards generally belong to the second category.

Yet their contribution should not be underestimated. They support efficiency, consistency, and trust.

And ultimately, they support better decision-making.

The first train that arrived in The Hague arrived at a station that is called Hollands Spoor.

This is only one of two major railway stations in this city. The other one is the central station.

Historically, they originated from two different railway networks, owned and operated by competing railway companies.

Nobody taking the train today thinks about that history, obviously. You hop on the train at either station, and you expect the train to take you to your destination. You expect the network to function as a network.

And that expectation rests on common standards and interoperability.

The same is increasingly true of data.

As Europe works to make supervision simpler, smarter and more structured, we should continue to invest in the foundations that make shared understanding possible.

XBRL has played an important role in building those foundations.

The challenge now is to ensure that it continues to do so in a world of larger datasets, closer cooperation and increasingly capable AI.

Because moving from data to insight ultimately depends not only on our ability to exchange information, but also on our ability to understand it in the same way. That is how we can keep Europe’s data, and European supervision, on the right track.

Thank you.