
Secrétariat général de la Commission bancaire Secrétariat général de la Commission bancaire IX European Banking Supervisors XBRL Workshop. Formulae in the European Finrep Taxonomy Paris, October 29, 2008 SGCB
• Target: Implementation of Finrep European Formulae • Background: Experience with formula 2004 and translation to 2008 version (software provider) • Test if and how each type of European formula can be processed with 2008 version. • Needs: European Business specification and technical harmonization Introduction
Summary 1 • French Background -3 different types of formulae -French Business Rules -From 2004 to 2008 Version 2 • European Formulae -Excel File for specification -Formulae types -Examples of architecture -Drawbacks to be considered 3 • Prototype For Implementation -Example of automatic formulae generation
1 •French Background Three types of French formulae • Simple aggregation all elements are primary elements that have the same context and unit ex: Calculation Link in all tables. several primary items in one formula with break down by dimension(ex: table 34 CL) • dimensional aggregation primary item(s) with same concept name, identifier, period, unit and dimension with each domain member one primary (ex: table 30B BR_49): «aggregator-compositor pattern» French particularity with one formula for each primary item • Simple formula addressing every element of the XBRL instance ex:between 100 and 200 formulae Table1.1 Form25
1 •French Background Business Rules -Missing facts: -2 taxonomies to manage different dates of reports (some CL are not triggered when no fact present witch is contrary to business people requirement who wanted an error to be raised) -Different acceptance radius if the bank belongs to « Grands groupes » or not. -Don’t need to deal with entity: one instance = one entity + one Taxonomy + one currency -Period: creation of an element « balance start » to differentiate starting and closing period with same date. -segment: no information to take into consideration for formulae in the segment tag
1 •French Background From 2004 to 2008 Version 2004: No modularization, no extensibility but readability, pure XPath -Software provider rules implemented by a member of our subcontractor team -Ex: BR9 Table7 -Equivalent of general variable in 2008 version 2008: Standardization, extensibility, modularity, general semantic -All CL, Aggregation Dimension and simple formulae implemented with Formula 2008 specifications (Software Provider) -Ex: BR9 Table7 -So far: Temporary solution for architecture 2008: Specification limitations -Inter-instances formulae problem: Resolution outside the scope of specification -Very difficult to implement chains (output of a formula becomes the input of another one) -No specification for output
2 •European Formulae Excel Files for specification Only business specification : Excel File provided with Finrep 1.3.0 Types of European formulae • Simple aggregation(around 500) ex: Calculation Link in all tables of 2008 Version automation is required - several primary items in one formula with break down by dimension(ex: table 30B BR_48) • dimensional aggregation(around 100) - one primary (ex: table 30B BR_49) • Simple formula(around 150) ex:Table1.1 Form25
Banco de España style guide 2 •European Formulae
2 •European Formulae • Trial implementation of French Formulae • • Simple aggregation: one file for each template(t-finrep-br-fr-2007-06-30- TableXX-formulae.xml) • Other formulae: one file for all formulae • Other solutions • Software vendor solutions: everything in one file
2 •European Formulae Drawbacks to be considered • Real Need of a shared architecture (best practice): all editors should be able to deal with modification of a shared architecture equivalent of Frta for XBRL 2.1 ex : -Spanish XSLT files and specification -Our Contractor implementation -Software provider implementation • Defining shared business rules at European level? -periodicity of reports for example -need of being generic for national extensions • Target: Firnrep with new ifrs • Formulae <> Business Intelligence = still superficial controls (no historical or complex statistical analysis) => not the same goal (control on a report) • Backward compatibility (CL, output) at least
3 •Prototype for implementation • java prototype: -500 calculations -Need of Mapping File between Excel cells and XBRL Taxonomies (Tool of our Software provider) Jérôme -Use of Bank of Spain xslt files -Java components : Poi, Jdom . • No European Codification: use of xbrl names
Conclusion • Basic Formulae can be implemented straight away • Need of business specification at European level • Need for standardization in the architecture for software developers