Avantage Reply was engaged to review and challenge a model used to comply with the US liquidity Coverage Ratio requirements.
The aims of the engagement were to review the documentation and implementation of the current model and build a challenger model. Based on this, Avantage Reply formulated a set of recommendations to increase the quality of the model and to ensure that regulatory expectations are met. In addition, the challenger model identified an alternative modelling approach that resolves some shortcomings of the current model.
Our client is a global financial services company with a strong presence in the banking sector, including brokerage, custody, fund management and asset servicing businesses.
Our client developed and implemented a model which separates operational deposits into core and excess balances and demonstrates that the core deposits are empirically linked to the operational services. By doing this, the model enabled the client to apply a lower (more favourable) run-off rate for the application of the US Liquidity Coverage Ratio (‘LCR’) requirements.
The model employed three conceptually different approaches depending on the line of business. These approaches reflected differences in data granularity and differences in the appropriateness of assumptions for various deposit sources. In addition, the documentation lacked clarity as to what assumptions actually applied to each deposit source.
Separately, the business wanted to increase the proportion of core balances within the operational deposits as these benefit from a more favourable treatment. However the model developers were unable to explain what actions should be taken to achieve this. For example, was it better to focus on increasing balances from specific clients or should it increase the number of clients to benefit from diversification effects?
The client engaged Avantage Reply to perform a review of the methodology, documentation and code to develop a challenger model. This exercise was performed separately from the ‘normal’ model validation and focused on addressing the lack of clarity in the documentation, understanding the drivers of the results and exploring alternative modelling approaches.
The approach adopted had two parts. The first part consisted of the review of the existing model and focused on:
A second part focused on developing a challenger model by exploring the use of alternative data and alternative statistical methods. The end result was two challenger models. One focused on using alternative data and demonstrated the appropriateness of the certain intermediary results. However this approach did not address the issue of having some intermediary results that transgressed one boundary condition. This was addressed by the other challenger model which demonstrated that by using an autoregressive model, existing assumptions could be dropped, and results obtained.
Avantage Reply provided the client with a report containing the analysis and all code files. A number of follow-up sessions were held with various stakeholders including the Model Developers, Business and IT to fully explain the recommendations and support their implementation. The recommendations implemented by the client were: (1) re-using sections of our report the model documentation and code were simplified, (2) possible business actions were identified, and (3) model enhancements were suggested that will be implemented once new data feeds are set-up (currently the new data was only available for a limited number of business lines). Finally, because of the data quality issues identified, improvements were made to the data governance and controls.