Data extraction

This posting focuses on the very first step that precedes any analysis of data and on the unique challenges it introduces.

Data extraction can be as simple as running a query when the organization has an integrated business intelligence system, but it can be much more complicated. Data extraction may involve retrieving data from mainframes, relational and non-relational databases, files of various formats, web services, packaged applications like ERP and CRM, and more.

Most BHCs are in relatively good shape when it comes to data extraction for DFAST- or CCAR-related analyses. Financial data is typically structured in tables of various formats. The structure may change over time or across units within the organization, name conventions may differ, they may be located at different places. At the end of the day, though, a table is a table. There is a structure, there are rules, and it is relatively easy to read and manipulate the contents. (Things are far more complex when the data is unstructured, but let’s not worry about that now).

While we strongly advocate using an integrated software solution to handle the various types of extraction scenarios across the entire organization, we are well equipped to handle a variety of system setups.

Currently, BHCs are responding to the regulator requirements and submit annual company-run stress testing reports. We believe that over time BHCs will see the benefits of stress testing and will better integrate stress testing into their decision making processes. This will require more frequent stress testing, and more frequent data extractions. Initially, stress testing will be done quarterly instead of annually, then monthly instead of quarterly, and eventually on an ongoing basis.

Data extraction will have to be done at a very high frequency – BHCs will invest in integrated software solutions because the benefits will outweigh the cost. Some of the larger BHCs are already investing substantial resources to develop such system (not only for the purpose of stress testing). But, most small and medium size BHCs are still lagging behind.

At ECON|Analysis, we recognize that BHCs may be at different stages of integrating such systems. We therefore customize our tools to work with the client’s existing systems. We’ll get the necessary data and prepare it for analysis. We will focus on the process of data preparation in our next posting.