XBRL Financial Reporting Taxonomy: Microstructure, Quality Evaluation And Improvement Method

XBRL had been practicing and developing for15years, after Charles Hoffman,CPA, applied a rich semantic technology-XML to financial reporting in1998andgradually formed a concept of extensible Business Reporting Language（XBRL）.Nowadays, the frontier of XBRL Financial Reporting Taxonomy has changed fromthe building and implementing the Taxonomy to evaluating and improving the Tax-onomy. Researches are rich on building and implementing the Taxonomy, while arerare on evaluating and improving the Taxonomy. So there are quite a few problemsneeded to be answered in these research fields.The first question is ‘What’s the Taxonomy’s microstructure?’. It is thefundmental problem in these research fields. Many scholars hold the opinions thatFinancial Information Element is the Taxonomy’s basic cell, however, the microstruc-ture of the Taxonomy had changed, accompanied by using ‘Dimension’ ModelingMethod. The answer of the problem not only contribute to the Financial InformationElement Theory, but also further the research direction on the assessment of theTaxonomy’s quality. The second question is ‘How to evaluate the Taxonomy’s Quali-ty?’. It is the core problem in these research fields. Many researchers evaluated thequality from the Integrity view, lack of creating and extension perspective. The objec-tive of evaluating the Taxonomy’s Quality is to find the pros and cons in different cre-ating and extension patterns, so as to afford qualitative and quantitative evidenceson making the Taxonomy better. The Third question is ‘How to improve the Taxono-my’s quality?’. It is the application problem in these research fields. Many research-ers had offered many policy proposals and implementation frameworks on im-provement of the General Taxonomy, however, whatever improvement, it is una-voidable to make a choice between Integrity and comparability. The Babel situationchanged by introducing the Industry-Level Taxonomy. While there is lack of effectivemethod on creating the Industry-Level Taxonomy. An operable method will contrib-ute to the improvement of the Taxonomy’s Quality, continuously improve the infor-mation exchanged between the financial and the economic entities, and meet the requirements of stakeholders on high-quality financial reporting. The dissertationfocuses on these three basic questions. Not only people from practice field such asTaxonomy creators, Instance creators, Financial information regulators, and even in-vestors but also theory researchers in the XBRL Financial Reporting focuses on thesequestions.Based on Microeconomics and Financial Accounting Theories, the dissertationsets evaluation and improvement of the Taxonomy’s Quality as topics, employs Set,Matrix, Probability and Mathematical Statistics theoretical analysis and empirical testmethod.The first research research question is ‘What’s the Taxonomy’s microstructure?’.This dissertation compares selection and collection methods of the Taxonomy by re-verse engineering, analyzes and compares the Taxonomy’s microstructure from Cre-ating and Extension view, and formally describes current Taxonomies.The second research question is ‘How to evaluate the Taxonomy’s Quality?’. Thestudy attempts to constructe criteria and indicators on the measurement of the Tax-onomy’s quality in the view of creation and expansion. On the aspect of CreatedQuality, this dissertation attempts to construct a Created Quality Standard, developsthe concrete Created Quality Models according to the Dimension Pattern and TuplePattern Finally, using Shanghai Stock Exchange’s Listed company Taxonomyrepresentativing Tuple Pattern, General Taxonomy representativing Dimension Pat-tern, the study selects information elements from the Notes to Financial Statementsas samples, measures and evaluates Created Efficiency, Semantic Integrity and Over-all Quality of different Created Patterns. On the aspect of Extended Quality, this dis-sertation attempts to construct a Extended Quality Standard, develops the concreteExtended Quality Models according to Industry Extension Pattern and Direct Exten-sion Pattern. Finally, using information elements from the Notes to Financial State-ments of34petro industry Listed companies as samples, the study measures, evalu-ates and robust checks the Integrity, Efficiency, and Comparability of different Exten-sion Patterns. The Third question is ‘How to improve the Taxonomy’s quality?’. The study at-tempts to offer operable theories and methods on creating the Industry-Level Tax-onomy and uses information elements from the Notes to Financial Statements of153manufacturing industry Listed companies as samples to build a manufacturing Indus-try-Level Taxonomy.The main research conclusions are as follows:1. The study puts forward that Taxonomy’s fundmental units vary from the Tupleto Dimension. Financial Information Element is the fundmental unit in the Tuple,while Construction Information Elements, including table, axis and line item, are thefundmental units in the Dimension, where are members in the axis and concepts inthe line item. Shadow Financial Information Elements are made of members andconcepts.2. Dimension Pattern is better than Tuple Pattern in the Created Quality. Indus-try Extension Pattern is better than Direct Extension Pattern in the Extended Quality.On the aspect of Created Quality, supposing efficiency and integrity are equal-ly important, unit cost are equal, and unit benefit are equal in the Information Ele-ment Space, the study finds that although Tuple Pattern is better than DimensionPattern in in the Semantic Integrity, Dimension Pattern is better than Tuple Pattern inthe Overall Quality and the Created Efficiency. On the aspect of Extended Quality,The study finds that there are Statistically and Economically significant differencesbetween two Extension Patterns’ Qualities of integrity, efficiency, and comparability.On average, Industry Extension Pattern is40percent better than Direct ExtensionPattern in comparability and dominant will not change with the narrowing scope ofinformation elements, reflecting the Comparability Measurement Model to listedcompanies is robust.3. The study puts forward a method on building Industry-Level Taxonomy, anddevelop a Manufacturing Industry-level Taxonomy. The artical develops a frequencymethod in choosing Information Elements on a basis of augmenting Information El-ement Space Theory, attempts to construct an Intuitive statistical method and aneconomical method on comparability utility theory. Using manufacturing industry Listed companies as samples, the study calculates related measurements, such asExtended Frequencies, Extended Densities, Query Cost of Regulative Ractor, AverageComparability, Modified Comparability, Cumulative Extended Element Quantity, andCumulative Extended Element Proportion etc. Secondly, the study reckons the bestExtended Frequency is66by the Comparability Utility Optimization Theory. Finally,the article chooses the Information Elements, whose Extended Frequency is greaterthan or equal to66, to develop a Manufacturing Industry-level Taxonomy.The innovation of this dissertation is in four aspects:1. The study deepens the Taxonomy’s Financial Information Element Theory andreconstructs the Taxonomy’s Information Element Space Theory. On the one hand,the dissertation formally describes the Taxonomy’s microstructure by the Set Theory,compares the Taxonomy’s Created Patterns and Extended Patterns, deepens theTaxonomy’s Financial Information Element Theory, and lays the theoretical founda-tion for evaluating the Taxonomy’s Quality. On the other hand, the study addsFrequeny to the Information Element Space for the first time, redefines the Space,constructs Frequeny-Density Space and Frequeny-Probability Density Space, extendsthe Information Element Space Theory from one dimension to multidimension, ex-tends Element Domain to Element-Frequeny Domain, Frequeny-Density Domain andFrequeny-Probability Density Domain, builds mapping relation functions among el-ement, frequency, density and probablity, and lays the theoretical foundation fornarrowing scope of information elements and developing Industry-level Taxonomy.2. The study constructs a measurement on the Taxonomy’s Created and Ex-tended Quality, and evaluates the Taxonomy’s Created and Extended Quality. On theaspect of Created Quality, based on cost-benefit and information completeness, thedissertation constructs a Created Quality Model, attemps to evaluate the Taxonomy’sCreated Quality in the direction of Created Efficiency, Semantic Integrity and OverallQuality, and affords quantitative evidences on creating the Taxonomy. On the aspectof Extended Quality, difference from evaluating the Taxonomy’s completeness by in-formation matching method in current issues, the dissertation introduces Frequencyto evaluating the Taxonomy’s Extended Quality, evaluates Taxonomy’s completeness by Cumulative Extended Element Quantity, evaluates Taxonomy’s efficiency by Cu-mulative reused Element Quantity, constructs Taxonomy’s comparability model byCumulative reused Element Quantity. Supposing completeness, efficiency and com-parability are equally important, the study constructs a Overall Extended QualityModel, attemps to evaluate the Taxonomy’s Extended Quality in the direction ofcompleteness, efficiency and comparability, and affords quantitative evidences onthe Taxonomy’s Extension.3. The study constructs a method on screening Taxonomy’s Information Ele-ments and building Industry-level Taxonomy by probability significance or compara-bility optimized utility. From the perspective of disclosure practice, the dissertationnarrows the Information Elements by Frequency method and develops the Indus-try-level Taxonomy. The study counts Frequencies of information elements from theNotes to Financial Statements, transforms Frequencies to Reused Densities and Ex-tended Densities, changes Densities to Probability Density, calculates the floor ofFrequency by Statistically and Economically significant, and chooses the InformationElements, greater than or equal to the floor, to develop the Industry-level Taxonomy.Statistically, the study directly narrows Information Elements by probability, whileeconomically, the study indirectly narrows Information Elements by the Comparabil-ity optimized utility, compensating for the lack of the method on selecting Infor-mation Elements.