01 May The days of the fully integrated data model are numbered
May 1, 2015 – By Pavel Yakovlev, Head of Solutions Team EMEA
The fully integrated data model for risk and finance is a concept that has been bandied around for years, but which remains as impractical and unsatisfactory as ever. Increasing numbers of firms are now turning their backs on this approach and looking for a more flexible alternative that is better suited to the fast-changing, heterogeneous regulatory environment in which they find themselves.
The fully integrated data model is often presented as a silver bullet for firms’ data management and reporting challenges. Advocates claim a single data layer can be used across risk and finance, and tell firms that if they feed their data into a fully integrated data model, it can then be used for all of their regulatory reporting in multiple jurisdictions, as well as their internal reporting.
It sounds wonderful, doesn’t it? Sadly, the shortcomings of the fully integrated data model become clear as soon as you attempt to put the theory into practice.
Consider the example of a firm that operates in multiple jurisdictions. If the firm uses a single integrated data model, it will need to update that data model across all internal operations every time a regulation changes in one of the jurisdictions in which it operates. This will involve expensive and time-consuming regression testing, and will impact the entire enterprise, even though the change is only relevant in one jurisdiction.
This is what happened earlier this year when the European Banking Authority (EBA) released version 2.3 of its XBRL taxonomy, which added European Union (EU) in the country field. Many firms were forced to make this change to the fully integrated data model they use for risk and finance in all jurisdictions, even though they would only use it in countries that are subject to EBA reporting.
The fully integrated data model is also totally impractical when attempting to manage the varying implementation timetables of different regulations. This can be seen in the example of the new Balance Sheet Item (BSI) and Monetary Financial Institution (MFI) Interest Rate (MIR) reporting requirements in Europe. Although these requirements have been agreed at a European level, they are being customized by local regulators, who are implementing them at different times. Those firms that use a fully integrated data model for everything will find all functions inconvenienced several times as the BSI/MIR requirements of local regulators are phased in at different times.
Using a fully integrated data model is not only inconvenient for firms, it is also a cause of concern for regulators, due to the possibility of correlation risk. If a firm uses a fully integrated data model that does not naturally fit its internal operations, it may fail to accurately report important risk factors to the regulators. If a large number of firms adopt the same fully integrated data model, the risk to the market is multiplied because all of their regulatory reports are likely to suffer from the same deficiencies, and the regulator may therefore be unable to identify significant changes in the market.
So what is the alternative? In response to the issues I have described above, we at AxiomSL have developed what we refer to as a loosely integrated taxonomy. This means that we provide separate taxonomies for the requirements of different regulators, including the EBA, European Central Bank (ECB) and domestic authorities, such as the De Nederlandsche Bank (DNB), Banco de España (BDE), the Swiss National Bank (SNB) and the Bank of England (BOE). We then highlight the data definitions that are common between different regulations and give users the option to implement a common business rule for these requirements.
The loosely integrated taxonomy gives firms complete flexibility. Whereas the single integrated data model forces firms to make difficult decisions upfront about how risk and finance will map to the same data model (which function will have to change and how etc.), the loosely integrated taxonomy allows firms to decide whether they want to leverage the commonality between different taxonomies immediately, at a later stage or never at all. The decision is entirely theirs. If a firm decides not to leverage commonality between taxonomies, this decision and the reasons for it will automatically be recorded and available if needed for audits.
The loosely integrated taxonomy gives firms the ability to respond quickly to change because all functions are not obliged to march in lockstep with one another. It also means that when one function needs to update its taxonomy, this will not have a knock-on effect on all other functions. On the other hand, the loosely integrated taxonomy gives firms the ability to avoid duplicate activities by allowing them to implement common business rules when there are overlaps between different requirements.
The fully integrated data model is a pipe dream that has never lived up to its big claims. Now is the time for firms to embrace an alternative that offers both flexibility and efficiency.
To discuss this article further, please contact:firstname.lastname@example.org