Locking down models via End User Computing Tool declaration

Problem

A traditional area of problems for legacy spreadsheets has been around governance: Who can see and use the model in some manner? How can the audience be assured that the model people are relying upon for some purpose has not changed?

Among numerous responsibilities, Controllers and their associated engineering teams at many financial institutions are tasked with certifying the valuations of specific transactions and usages of their firms’ capital. These usages may face formal, additional scrutiny by state and federal regulators. Mistakes can be costly, not just to the firm’s earnings but possibly resulting in steep fines and penalties. Because their work needs to be independent of other systems, legacy spreadsheets have been the go-to tool of choice.

Solution

In one diverse set of such situations, an ecosystem of ClearFactr models was deployed to replace these legacy spreadsheets. Many of the models, when published to a particular audience, were published using ClearFactr’s “EUCT” workflow:

1. A model is published to a particular team which often included members with API credentials.
2. Prior to becoming visible to the entire team, a chosen Approver – from a subset of team members with this particular role/permission -- must officially review the model and attest that they have done so. This workflow is performed entirely within the ClearFactr platform, with emails being sent to the various parties at each stage.
3. Upon approval, the model is designated as “published” , and the entire team gains access at that time.
4. Fine-grained permissioning, per team member, can be maintained via the Model Engagement panel.
5. Most notably, once published in this mode, the system prevents any formulaic cell in the model from being changed by anyone. To make such a change, the model must be formally unpublished first, then re-approved during a new publishing process.

Much of the recurring valuation workflows were automated via ClearFactr’s API:

1. Inputs to the model would be updated via one or more of the family of cell-changing endpoints. At the extreme, an entire tab’s content could be replaced with a single endpoint call.
2. Other software would pull computed values from these models, treating them as datasources for additional processing and/or reporting downstream.

Key Results and Benefits

1. Thousands of inconsistent valuation files covering hundreds of investments were replaced with a smaller, yet more powerful ecosystem of ClearFactr models.
2. Consumers of these models gained dramatically higher confidence in the quality of these models as datasources, knowing that they were managed by a formal, system-directed workflow with rigid change management.
3. Consumers of the models also understood the models more readily, due to the explanatory features of the platform and also because the models were increasingly standardized into families of solutions. The system encouraged and rewarded a more thoughtful modeling process.
4. Team productivity soared -- the amount of regulatory capital that could be certified by the same size team doubled.
5. When discussing the overall impact of ClearFactr on their work, one team member remarked: “We used to start off every meeting wasting valuable time simply aligning on which version of which Excel file we were here to talk about. Now we can actually be Controllers again, discussing key business logic and assumptions instead.”