Digital Financial Reporting

This blog post contrasts two different approaches to constructing XBRL taxonomies: bottom up and top down.

I don't know if you want to call these techniques, or strategies, or philosophies, or architectures.  And I don't know if "bottom up" and "top down" are the best choice of terms, but I will go with those terms now until I find something better.

By bottom up I mean that concepts are defined and relations between concepts are defined but the sets of concepts and relations are not explicitly defined.  The US GAAP XBRL Taxonomy and the IFRS XBRL Taxonomy both use this bottom up approach.

By top down I mean that the actual sets themselves are explicitly and uniquely defined and given names, and then the concepts and relations between concepts are defined.  The XASB reporting scheme uses this approach as does the old COREP and FINREP taxonomies.

Each approach has its pros and cons.  It is not the case that one approach is good and the other approach is bad; they are simply different approaches.  Most people creating taxonomies are completely unconscious that they even have a choice.

The primary advantage of the top down approach is that you can query reported information by the name of the set of information because each set of information is uniquely named and identified by a specific hypercube that is only used to report that specific information.  For example, if you open the XASB reporting scheme sample report that I created and look at the hypercubes (also called [Table]s) you will see that they are all unique.  If you contrast that to the Microsoft Corporation XBRL-based report submitted to the SEC and you notice how Microsoft uses the hypercube or [Table] us-gaap:StatementTable to represent many different disclosures; then you think about what it would take to query the balance sheet then you will understand the difference between the top down and bottom up approaches.

Now, you can "convert" a bottom up approach to a top down approach.  I have done that by creating the necessary machine-readable metadata.  That means that I can query bottom up or top down taxonomies.

Here are examples of both a bottom up taxonomy and a top down taxonomy where I have done exactly that, successfully queried sets of the report.  Notice in the two examples the list of "disclosures" and the business rules used to find the disclosure.  The key to understanding the difference is the composition of the business rules that are used to identify each disclosure within a financial report:

• Bottom up: XBRL-based financial reporting by public companies to the SEC. (This video uses this same summary of disclosures.)
• Top down: XBRL-based financial reporting per the XASB reporting scheme.

One "advantage" of the bottom up approach is that you don't have to explicitly define everything in the taxonomy in terms of the disclosure it represents.  This is not necessarily an advantage because if you make a mistake and leave something out, that error is not apparent.  To do this though, you have to understand 100% of the disclosures that will be provided.

Whereas the top down approach requires you to understand and explicitly provide for each disclosure up front and you simply build those sets one-by-one and piece-by-piece until you have the entire set of explicit disclosures provided for.

Another example of top down as contrast to bottom up is the extraction of information from reports: (pay particular attention to the mapping rules and impute rules provided in the VBA code of the Excel extraction tool)

• Bottom up: Extracting information from one reporting style related to XBRL-based reporting to the SEC.
• Top down: Extraction from one reporting style related to the XBRL-based reporting under the XASB reporting scheme.