230 likes | 414 Views
Hiding the Angle Brackets. Rendering XBRL for compliance professionals and regulators Lucian Holland 27 April 2005. Introductions. About DecisionSoft Leading supplier of XBRL technology and consulting to regulators worldwide Industry standard True North Validator About me
E N D
Hiding the Angle Brackets Rendering XBRL for compliance professionals and regulators Lucian Holland 27 April 2005
Introductions • About DecisionSoft • Leading supplier of XBRL technology and consulting to regulators worldwide • Industry standard True North Validator • About me • Technical Architect responsible for True North • Regular contributor to XBRL and XML conferences
Presentation overview • The rendering problem • Requirements • Challenges • Look at available solutions • Focus on one promising possibility
The CFO’s Dilemma This is some very tiny text that is supposed to be part of the report that is on display. <usfr-pt:MarketableSecurities AvailableSecuritiesCurrentNon current precision=“INF” contextRef =“Inst2002” unitRef=“USDollar”> 23445182</usfr-pt:MarketableSecurities AvailableSecuritiesCurrentNoncurrent> <usfr-fst:LoansHeldSale precision=“INF” contextRef=“Inst2002” unitRef=“USDollar” >7747793</usfr-fst:LoansHeldSale><usfr-fst: LoansLoansHeldPortfolio precision=“INF” contextRef="Inst2002“ unitRef="USDollar"> 73167935</usfr-fst:LoansLoansHeldPortfolio> <usfr-fst:NetLoansAllowanceLoanLeaseLosses precision=“INF” contextRef=“Inst2002” unitRef= “USDollar”>930114</usfr-fst:NetLoansAllowanceLoan LeaseLosses> ??**£$&%!
The need for readability • The CFO wants to read what he’s sending out • Regulators want to cross-check reports flagged for closer inspection • Analysts and investors want to see the entity’s own presentation as well as the new analyses made possible by XBRL • Key driver for adoption – people must be able to “see inside” the XBRL
Requirements for rendering • Comprehensible to the non-XBRL-literate • XBRL is not human-readable • XBRL does things differently to traditional presentations • Needs a robust “sign-off” mechanism • Get to the point where the XBRL can be audited? • Transparent mapping between rendering and XBRL • Information mustn’t be able to change after sign-off • Signing must be integrated to be of real use • Portability • No specialist software at point of consumption • Needs to be lightweight
Challenges • Dimensions • Rendering patterns • Segmentation of information
Dimensions • XBRL supports multi-dimensional information • Will soon support even more with new Dimensional Taxonomies work • Which dimensions should you use for display? • Facts can be positioned on 3,4,5 (…100?) dimensions at once – which ones should be used? • Dimensions built-in to existing report layouts • XBRL will often be generated from such layouts (e.g. Excel); dimensional structure gets “lost in translation” • Taxonomy authors have minimal control • Almost no interaction possible between semantic structure of taxonomy and dimensions expressed in reports.
Period: 2004 Segment: US Total Scenario: Actual Period: 2004 Segment: UK Total Scenario: Actual Period: 2005 Segment: US Total Scenario: Actual Period: 2005 Segment: UK Total Scenario: Actual Dimensions: an example World-wide overview UK report 2000 1000 200 100
Movement analysis • A specific problem with dates in XBRL • Relationship between fact and date changes semantics: beginning vs.end • No way to make this explicit in standard XBRL • Perhaps should be fixed at a specification level, but unlikely to happen soon • Important for rendering • Will require tool support, along with more metadata
periodType: Instant Movement analysis Taxonomy Balance Change in balance Report periodType: Duration Movement analysis: an example Date: 2004-04-01 Scenario: Actual Date: 2004-10-01 Scenario: Actual Movement analysis Balance=100 Date: 2005-04-01 Scenario: Actual Balance=200 Date: 2004-10-01 Scenario: Projected Balance=250 Balance=200 Date: 2005-04-01 Scenario: Projected Balance=100 Change=100 Period: H1 2004-5 Scenario: Actual Change=50 Change=-100 Period: H2 2004-5 Scenario: Actual Period: H2 2004-5 Scenario: Projected
2004 Actual Movement analysis 2004-04-01 – 2004-10-01 Starting Balance 100 Profits 100 Ending Balance 200 2004-10-01 – 2005-04-01 Starting Balance 200 Profits 50 Ending Balance 250 2004 H2 Projected 2004-10-01 – 2005-04-01 Starting Balance 100 Profits 100 Ending Balance 200 Movement analysis example continued Date: 2004-04-01 Scenario: Actual Date: 2004-10-01 Scenario: Actual Balance=100 Date: 2005-04-01 Scenario: Actual Balance=200 Balance=250 Date: 2004-10-01 Scenario: Projected Balance=200 Balance=100 Date: 2005-04-01 Scenario: Projected Change=100 Change=50 Period: H1 2004-5 Scenario: Actual Change=-100 Period: H2 2004-5 Scenario: Actual Period: H2 2004-5 Scenario: Projected
Segmentation • Presentation linkbase only gives a hierarchy • No indication of how to split into tables/sections • As discussed, no link to date, segment or scenario dimensions • Unclear what to do when information is missing • Already a problem for calculations in US-GAAP sample preparation • Will be problematic for rendering as well – how to present a hierarchical structure when some parts may be missing?
Presentation overview • The rendering problem • Requirements • Challenges • Look at available solutions • Focus on one promising possibility
The shape of a solution • Must handle complex processing – simple transformations inadequate • Will need significant metadata: • “Patterns” – identify standard ways of expressing information and define renderings (e.g. describe the XBRL “shape” of a movement analysis) • “Hints” – allow authors of taxonomies or instances to specify rendering in more detail (e.g. identify specific elements used for movement analysis) • Must be possible for non-technical domain experts to define and shape this metadata
Spreadsheet 1 2 3 A 32423 23234 22 B 3789 356 34 C 4799 4394 26 D 338 5799 43 Report Possible solutions: Rekeying • Requires no technology investment • Unreliable • Expensive • Huge maintenance cost • Assumes that XBRL is not the primary format
<HTML> Processor 1 2 3 A 32423 23234 22 B 3789 356 34 C 4799 4394 26 D 338 5799 43 Report Possible solutions: Standalone HTML • Portable, open standard • Signing possible, but not built-in • Not an integrated solution, lacks a lot of features • Hard for domain experts to define format • Every report must pass through renderer Distributable Package
Web Browser <HTML> 1 2 3 A 32423 23234 22 B 3789 356 34 C 4799 4394 26 D 338 5799 43 <XSLT> Report Possible solutions: XSLT • Widely available, standard and light weight • Rendering can be generated by end user • Can’t sign final rendering • Poor support/performance for complex processing • Nightmarishly difficult to support extension taxonomies • Specialist technical knowledge required to write • Works best for “closed form” solutions Distributable Package
Presentation overview • The rendering problem • Requirements • Challenges • Look at available solutions • Focus on one promising possibility
The real answer: PDF • “PDF – isn’t that just a static rendering format?” • Can contain a wide variety of machine-readable data including XML • Dynamic layouts that restructure based on XML data • Interactive controls • Sophisticated document management including element-level digital signing
Report Report PDF PDF: Encapsulating an XBRL instance • Rich client, ubiquitously available • Already a standard medium for accounts presentation • Rendering driven directly from the data • Signing the rendered form signs the data • Can support complex document workflows • Sophisticated design environments
Conclusions • Rendering XBRL poses a lot of challenges technically, logically and on a business level • The obvious solutions aren’t necessarily the right ones! • In general terms, the way forward is to look for “patterns” of standard usage and build rendering support for them • May even lead to a formal specification in the long run • However the “shape” of the output is defined, PDF has the edge as a delivery mechanism: power, flexibility and ubiquitous deployment
< /> Thanks for listening!