E-government
Article from BaFin's 2017 annual report
E-purchasing and e-invoicing
In 2017, BaFin set up an electronic procurement platform and is using it increasingly to conduct procurement procedures. It enables BaFin to handle all Europe-wide procurement procedures electronically well in advance of the implementation date1 prescribed by the legislation.
BaFin is also establishing a largely standardised and efficient workflow-based system for electronically supported invoice processing. This means that it is complying with the statutory requirements of the Directive on electronic invoicing in public procurement.2
Automated data exchange
BaFin uses the XBRL standard for the automatic exchange of business and financial data with supervised entities (see info box).
Definition:XBRL
XBRL stands for eXtensible Business Reporting Language and is an XML-based (Extensible Markup Language)3 standard which automates and standardises the exchange of data between two parties. The data are specified in taxonomies. A taxonomy is a catalogue in which elements are defined which can make up an XBRL data package. In addition to the actual format, such as the type of data, the definition also includes rules for checking data consistency. Thus, media discontinuity and the manual capture of financial data can be avoided.
BaFin applied this standard for the first time with the entry into force of the Solvency II Directive on 1 January 2016. The European Insurance and Occupational Pensions Authority (EIOPA) had developed a corresponding XBRL taxonomy for this purpose, which is used both by the insurance undertakings being supervised and by the national supervisory authorities.
This results in a high number of valid notifications from the beginning. The automated processing at BaFin generally takes a few minutes. These processing times have proved to be very good in a European comparison.
BaFin applied this standard for the first time with the entry into force of the Solvency II Directive on 1 January 2016. The European Insurance and Occupational Pensions Authority (EIOPA) had developed a corresponding XBRL taxonomy for this purpose, which is used both by the insurance undertakings being supervised and by the national supervisory authorities.
This results in a high number of valid notifications from the beginning. The automated processing at BaFin generally takes a few minutes. These processing times have proved to be very good in a European comparison.
BaFin is currently giving consideration to its own national taxonomy and is also making preparations for further deliveries of data on an XBRL basis.
Analytic data warehouse
The availability of large quantities of data (big data) requires new approaches to enable appropriate information to be extracted from them. BaFin has therefore introduced an analytic data warehouse (ADW), which is a scalable high-end in-memory database architecture. The ADW enables data to be analysed and evaluated almost in real time. Additionally, BaFin has begun the use of a visualisation software programme and initial artificial intelligence (AI) techniques.
Project ALMA
March 2017 saw the start of the interdisciplinary project known as the automated alarm and market monitoring system (ALMA). This uses the ADW as well as the visualisation software and initial AI techniques. For the automated identification of cases of insider trading in securities, ALMA extracts price-relevant information from ad hoc announcements and performs a variety of pattern recognition functions.
The ALMA project does not integrate new requirements in the context of a traditional concept phase. All of the participants engaged in this project within BaFin form part of a learning process in handling the data and are working out requirements, proto- and pseudocodes and indicators on an interdisciplinary basis.
The information required is described formally in a structured query language (SQL) so that it is located by the database system of the ADW in the gigantic volume of data in an optimised reaction time. For the data analysts in the relevant divisions, these indicators4 and SQLs form the basis for learning about ordinary trading activities from the system. This enables irregularities to be identified automatically and to be used to support the decision-making process.
ALMA integrates these algorithms and techniques in a single, intuitive visual interface for users at BaFin. The data transmitted to BaFin are stored in the ADW for this purpose via a number of process stages. The volume of data will be multiplied further in future as a result of new regulations coming into force.
Footnotes:
- 1 Directive (EU) 2014/24/EU, OJ EU L 94/65.
- 2 Directive (EU) 2014/55/EU, OJ EU L 133/1.
- 3 XML is a machine-readable language for presenting hierarchically structured data in the form of a text file.
- 4 The indicators include, for example, events such as date, actor and price and arithmetical variables such as mean values and significances.
