400,000 – that is the vast number of regulatory reporting rules and data elements that institutions have to implement and submit individually. This is mainly because, over time, manifold heterogeneously defined reporting obligations have emerged from a multitude of banking statistics, prudential and resolution reporting requirements.

For institutions, reporting has become increasingly burdensome. The existing approach is inefficient, makes little use of digital technologies, and requires long implementation cycles. Supervisors are also suffering as a result of this complexity and are endeavouring to increase analysis and response capabilities. “We need a state-of-the-art, fully digital reporting system because, as supervisors with a preventive approach, we are increasingly dependent on up-to-date, granular information,” explained Raimund Röseler, Chief Executive Director of Banking Supervision at BaFin. This is why – together with the Deutsche Bundesbank and five credit institutions, in addition to service providers and industry associations – BaFin has carried out a feasibility study on the reporting system of the future. In conducting the study, BaFin was supported by the consultancy firm Accenture (see info box).

Reducing the burden for institutions and improved capabilities for supervisory analyses

The aim of BaFin's study was to develop options for improving the analytical capabilities of supervisors while at the same time alleviating the reporting burden for banks, and to test these approaches using prototypes with pilot banks and service providers (see Figure 1). The project participants took into account recent European initiatives, such as the Eurosystem’s Integrated Reporting Framework (IReF), the Banks’ Integrated Reporting Dictionary (BIRD) and the feasibility study on an integrated reporting system produced by the European Banking Authority (EBA). The result is unequivocal: the banking supervisory reporting system must be promptly redesigned. It became also clear that such a change could be feasible in the future.. The participants in the study developed a vision that could achieve the greatest possible benefit for all parties.

Figure 1: Goals of BaFin study

© BaFin

Win-win vision developed

Various scenarios were examined as part of the study. The participants developed a “win-win vision” and validated it with the help of a digital prototype. Röseler is convinced of this vision: “This would bring us closer to our goal of being able to access, at any time, the data we as supervisors need. And at the same time, we would make reporting less burdensome for institutions.”

The vision is based on a common mixed granular data model that encompasses reporting requirements from all reporting domains (see Figure 2). This means that, as well as granular data, the model also holds some aggregated data that cannot be expressed on granular level. Hence granular data would only be reported once. Another component of the vision is a machine-readable rule set for data quality checks and data point aggregation rules which is built on the common data model. This “digital foundation” should be established on EU level and ideally in cooperation with the banking industry.

Figure 2: Vision

© BaFin

In this vision, institutions transform their heterogeneous, mostly granular data according to the common data model (BIRD X). The model is “mixed granular” because not all requirements can be reasonably expressed at a granular level, and certain key measures would continue to be delivered in aggregate form as “anchor values”. Although providing granular data and monitoring the quality on granular level would require more effort by institutions, the approach would also allow well over 75 per cent of existing complex data point aggregates to be generated centrally, thus providing significant relief for institutions.

Benefits for institutions and supervisors

A common data model would significantly improve the interpretation and implementation of reporting requirements and the interaction between supervisors and institutions. Additionally, it could provide the basis for a granular, linkable data set that enables redundancy-free submission. Before submitting their data, institutions could – within their local systems – make use of common data quality checks straight from the central machine-readable rule set – a significant synergy since they would no longer have to interpret reporting requirements individually.

Due to the higher data granularity, supervisors could conduct analyses with greater flexibility, expand their analytical capabilities and significantly reduce ad hoc requests by combining and evaluating data points by themselves. Only a central data aggregation can create the aggregation lineage as a direct link between granular data and data point aggregates. This enables seamless transitions between granular and aggregated views via drill down and drill up. Drilling down means exploring the underlying components at a more granular level from the aggregated level. Drilling up means understanding which granular data element contributes to which data aggregates.

On the one hand, anchor values would safeguard the aggregation process, and on the other hand, they would also ensure the institutions’ responsibility for key supervisory ratios. In the common data model, authorities could additionally leverage synergies by reusing and increasingly sharing existing data definitions, referred to as “define once” synergies.

The processes and advantages of the described vision were tested in a simplified prototype based on the logical BIRD data model. A cost-benefit analysis indicated that the vision developed was the most economically viable scenario.

European integration as a precondition for implementing the vision

For Röseler, the key step now is to integrate the results into the European process. “We are going to advocate for our project,” he announced. The vision already borrows from and further develops approaches from the IReF and the EBA feasibility study – for example by adding machine-readable rules and anchor values, and making templates technically redundant, since the required data are generated through a central data aggregation.

Figure 3: Scenario for further discussion of the vision

© BaFin

The implementation of the IReF announced for 2024 to 2027 could form the first stage (minimum viable product) for the vision developed in the feasibility study. To leverage its full potential, however the current legal framework would have to be adapted and the European initiatives would have to agree on a common vision and data model. The Joint Reporting Committee (JRC) announced by the EBA could lay the governance foundations for this. The BIRD could form the nucleus to develop the common data model. Now is an ideal time window to jointly redesign the reporting system on a sustainable common digital foundation (see Figure 3). “We need to use this time period – in our own interests and in the interests of the institutions,” emphasised Röseler.