Large Bank Examination Workshop February 2026

The Bank is clearly adhering to industry best practices and regulatory guidance by working with

qualified vendors (Mountain View and the service bureau) to develop decay rate and beta

assumptions that are derived from internal institution-specific data. However, the beta assumptions

date from analysis conducted in 2013 and Bank management is encouraged to determine if updated

assumptions would be more appropriate going forward.

II – Model Theory and Model Code and Mechanics

As referenced above, we leveraged our experience in serving as the resource engaged by the

Model developer and vendor to certify the 2012, 2014 and 2017 versions of the ALMTech ALM

Models. An extensive evaluation of the Model’s net interest income (NII) simulation, economic value

of equity (EVE), and re-pricing Gap methodologies was conducted during this evaluation. Based on

this review, Strategic Financial Solutions certified that the ALMTech Model possesses the necessary

theoretical components and operational capabilities to accurately assess the interest rate risk

sensitivity of a depository financial institution’s earnings and economic capital. This certification

was conducted in the context of universally accepted regulatory requirements as well as widely

recognized best practices in risk management. Institutions that utilize the Model by providing

financial data that accurately reflects the current composition of assets and liabilities along with

reasonable behavioral assumptions regarding historical and future customer behaviors can have

confidence that the Model can be used for its intended purpose. The Model was most recently

certified by MountainView LLC, on 4/30/2019 (see workpaper “2019 Statement of Model

Certification.pdf”), which affirmed that the ALMTech model possesses the innate design elements,

functionality, and user features required to accurately analyze IRR in financial institution balance

sheets.

V - Reporting

After completing a review of the computational methods, the Model output reports as well as the

report generating processes were addressed. The sufficiency of the reports themselves were

6

Page 18 of 39