Data & technology as it relates to model governance

Data & technology as it relates to model governance

Jeremy, you will be presenting at Stress Testing USA 2016. You will provide insights into how data and technology relates to model governance. What key points do you expect to cover at the event?

There are four key areas:

  • The requirements from regulators to demonstrate transparency and governance around the models
  • The inefficiencies due to a lack of transparency of the ecosystem that surrounds models
  • Focus on the tools feeding the models
  • Prevalence and impact of End User Computing (EUC) in the model governance environment

What impact does poor data have on financial modelling?

Up until now, the regulators have focussed their efforts on ensuring model governance (e.g. SR11-7, CCAR & DFAST), but recently, there is more emphasis from the regulators for banks to manage the tools that feed the models – with a focus on the EUC applications such as spreadsheets, databases and other financial modelling tools.

If the data in the EUC applications is not accurate, then it invalidates all the work that is undertaken in the model. Additionally, if organisations don’t have confidence in their data, then they devise a number of manual processes to understand where there may be inaccuracies and the corresponding impact on the output of the models. This is an enormous amount of wasted manual effort. Not only are they attempting to identify the poor data feeding these tools, they are also manually double checking to ensure the accuracy of the data as it is transformed within the spreadsheet. The process is hugely inefficient.

In your experience, what have you found are the consequences of a bank having poor data? What must be done to improve?

Poor data is an issue for banks. Only recently, a paper published by the Adam Smith Institute declares that the Bank of England’s stress testing program is plagued by a series of ‘fatal flaws’.  With most stress test exercises reliant or completely run via spreadsheet-based models in banks and financial institutions, the report categorically calls out the problem of poor data causing errors in calculations.

From my personal experience in talking to banks, they often have to undergo many fire-drills to try and understand where data issues are coming from. Banks need transparency around how data is created and where the transformations in the data and models are occurring in order to establish processes and controls to ensure good quality data. Establishing more transparency reduces the need for manual reconciliations and enables credible demonstration of a robust governance framework around the models and tools, which in turn gives greater assurance on the accuracy of outputs.

You will discuss the use of technology to support model risk governance. How does ClusterSeven assist banks with their model risk governance? ­­­

Our offering is primarily focused at the tool level. Our solutions enable banks to understand and control the entire data ecosystem that surrounds the models. This capability is extremely important for financial institutions as even a single error in one spreadsheet can proliferate right across the landscape, feed data into the various models producing inaccurate outputs.

The ClusterSeven solutions enable banks to exhaustively undertake both qualitative and quantitative analysis to determine where the data is coming from and into which models it’s going. Banks have complete visibility of all tools and a holistic view of the complex web of data flows, on an ongoing basis.

The solution enables banks to set up change management processes and control mechanisms to ensure that the integrity of the data is maintained. Furthermore, the automated processes facilitate re-attestation of the models and tools as banks regularly re-evaluate that the models are indeed working as desired by the organisation.

What key challenges do you believe stress testing professionals will face over the next 6 – 12 months?

Given the regulatory landscape, financial institutions will need to continuously adapt to changing and indeed growing reporting demands and requirements.  To deliver against such demands, the ability to quickly create transparent processes and establish the requisite controls will be critical, so that they can credibly demonstrate compliance to the regulators.

The Business Case for End User Computing: A ClusterSeven Whitepaper by Henry Umney, VP Sales, ClusterSeven

Spreadsheets and other forms of end user computing (EUC)[1] applications occupy the grey zone of enterprise technology. This whitepaper discusses why EUC applications exist in modern businesses, extending the discussion of risk, reputational, cost and regulatory issues recently reported by Chartis[2], before considering the returns to be gained by the use of ClusterSeven technology to manage the EUC landscape.

To download a copy please click here

Our 6th annual Risk Americas 2017 Convention will be taking place on May 23-24 and will feature a dedicated track on Stress Testing and Model Risk. Register before the agenda is announced for the heavily reduced rate of only $999.