PRIVILEGED AND CONFIDENTIAL ATTORNEY WORK PRODUCT — PREPARED IN ANTICIPATION OF LITIGATION
This memorandum constitutes attorney work product prepared at the direction of counsel in anticipation of litigation. It is protected by the attorney work-product doctrine, the attorney-client privilege, and the Kovel doctrine. This document and its contents are not to be disclosed to any person or entity outside of Calderwood & Harkness LLP and Brindoval Analytic Partners LLC without the prior written authorization of Victoria Chambers, Partner, Calderwood & Harkness LLP.
MEMORANDUM
December 22, 2023
FROM: Calderwood & Harkness LLP Victoria Chambers, Partner Jasmine Thornton, Lead Associate
Brindoval Analytic Partners LLC Dr. Franklin Osei, CPA, CFF, Engagement Lead
TO: File — SEC Investigation, Belmont Ridge Bank National Association, File No. HO-14287
RE: Analysis of Belmont Ridge Bank NA's CECL (ASC 326) Methodology — Allowance for Credit Losses, BridgeFin Digital Lending Portfolio
MATTER NO.: SC-MAT-2023-4172
I. Executive Summary
This joint attorney work-product memorandum has been prepared by Calderwood & Harkness LLP ("Calderwood & Harkness" or "Firm") and Brindoval Analytic Partners LLC ("Brindoval") to analyze the reasonableness of Belmont Ridge Bank National Association's ("BRB" or the "Bank") allowance for credit losses ("ACL") methodology as applied to the BridgeFin digital lending portfolio (the "BridgeFin Portfolio" or "Portfolio"). This analysis has been undertaken in connection with the investigation initiated by the Division of Enforcement of the United States Securities and Exchange Commission ("SEC" or the "Commission"), designated as File No. HO-14287 and conducted pursuant to Formal Order of Investigation No. EN-2023-0947, issued on October 12, 2023.
The scope of this memorandum encompasses a comprehensive evaluation of BRB's current expected credit losses ("CECL") model implemented under Accounting Standards Codification Topic 326 — Financial Instruments: Credit Losses ("ASC 326"), with particular focus on the following discrete areas of the BridgeFin Portfolio: (i) vintage-cohort treatment and classification methodology; (ii) re-aging and loan modification practices and their integration into the CECL model; (iii) loss-rate modeling assumptions and calibration; and (iv) qualitative overlay adjustments applied to the BridgeFin consumer personal-loan and micro-business loan sub-segments. The evaluation period centers on the fiscal year ended December 31, 2022 ("FY 2022"), with supplemental consideration of the quarterly periods ended March 31, 2023 and June 30, 2023.
Key Conclusion. Based upon the extensive analysis described in this memorandum and detailed further in the supporting appendices, it is the joint assessment of Calderwood & Harkness and Brindoval that BRB's reported ACL of $143.5 million — representing a coverage ratio of 3.50% against $4.1 billion in outstanding BridgeFin loans as of December 31, 2022 — was within the range of reasonable estimates under ASC 326 and was consistent with the CECL framework's express allowance for the exercise of informed management judgment. BRB's CECL model was developed through a rigorous process involving quantitative risk analysts, reviewed and approved by senior management, audited by the Bank's independent registered public accounting firm, and overseen by the Bank's Audit Committee. The methodology employed by BRB, including its treatment of re-aged and modified loans, is neither prohibited by ASC 326 nor inconsistent with prevailing industry practice.
The SEC Division of Enforcement Staff (the "Staff") has communicated a preliminary view that BRB's ACL should have been approximately $180.8 million, reflecting a coverage ratio of approximately 4.41% of outstanding BridgeFin loans. This preliminary view rests principally on an alternative treatment of re-aged and modified loans within the vintage-cohort framework — specifically, the Staff's contention that the 12,847 loans modified during FY 2022 (with an aggregate outstanding balance of $236.4 million) should have been retained in their original vintage delinquency buckets for loss-rate estimation purposes, rather than reclassified to "current" status in a new or adjusted performance cohort. While this alternative treatment is analytically defensible as one point within a range of permissible approaches, it is not the only permissible approach, and its adoption does not render BRB's chosen methodology unreasonable or BRB's disclosures materially misleading.
The potential understatement identified by the Staff is approximately $37.3 million ($180.8 million minus $143.5 million). While this difference exceeds BRB's internal materiality threshold of $30.6 million — calculated as 5% of FY 2022 pre-tax income of $612 million — by approximately $6.7 million, materiality under the SEC's own guidance (Staff Accounting Bulletin No. 99) requires both quantitative and qualitative assessment. Brindoval's analysis demonstrates that the $37.3 million difference: (a) is attributable to a discrete, documented, and consistently applied methodological judgment regarding the vintage-cohort classification of modified loans, not to a systemic error, data manipulation, or intentional understatement; (b) does not mask a change in earnings or earnings trends; (c) did not cause BRB to miss analyst consensus estimates; and (d) does not convert a loss to income or vice versa. Moreover, Brindoval's independent reconstruction of the CECL model demonstrates that both BRB's reported ACL of $143.5 million and the Staff's alternative of $180.8 million fall within the range of reasonable estimates (approximately $138 million to $185 million) produced by varying reasonable assumptions across the model's key parameters.
For these reasons, as detailed in the sections that follow, it is our conclusion that the facts and circumstances do not support a finding that BRB's ACL was materially misstated or that BRB's periodic filings with the Commission were materially misleading.
II. Background
A. Belmont Ridge Bank NA — Overview
Belmont Ridge Bank National Association is a national banking association organized and existing under the laws of the United States, operating pursuant to Charter No. 24891 issued by the Office of the Comptroller of the Currency ("OCC"). The Bank is headquartered at 200 Park Avenue South, 38th Floor, New York, New York 10003. BRB operates as a wholly owned subsidiary of Belmont Ridge Financial Corporation, a Delaware corporation, which serves as the Bank's holding company and is the registrant for purposes of the Bank's SEC filings.
BRB's common stock is publicly traded on the Nasdaq Stock Market under the ticker symbol "BRDG." As of September 30, 2023, BRB's approximate market capitalization was $4.7 billion, with approximately 76.74 million common shares outstanding.
The following key financial data summarize BRB's consolidated financial position as of and for the fiscal year ended December 31, 2022, as reported in BRB's Annual Report on Form 10-K filed with the Commission on February 28, 2023:
BRB's senior leadership relevant to this memorandum includes the following individuals:
• Chief Executive Officer: Marcus R. Ellingham
• Chief Financial Officer: Patricia "Trish" Navarrete
• General Counsel and Chief Legal Officer: Derek Sung
• Chief Risk Officer: Annette Kowalski
• Chief Accounting Officer: Raymond Hu
• Audit Committee Chair (Board of Directors): Gwendolyn Prescott
BRB is subject to comprehensive supervision and examination by the OCC as its primary federal banking regulator. The Bank's most recent OCC examination, completed as of September 2023, resulted in a composite CAMELS rating of 2 (Satisfactory). This rating reflects the OCC's overall assessment of the Bank's capital adequacy, asset quality, management capability, earnings strength, liquidity position, and sensitivity to market risk. Notably, the September 2023 examination did not cite any material deficiency in BRB's ACL methodology or in its credit-risk management practices with respect to the BridgeFin Portfolio.
BRB adopted ASC 326 (the CECL standard) effective January 1, 2020, the mandatory adoption date for SEC-reporting entities classified as large accelerated filers. The adoption was accomplished through a modified retrospective transition approach, with a cumulative-effect adjustment to retained earnings recorded as of the adoption date, consistent with the transition provisions of ASC 326-10-65-1.
B. BridgeFin Division — Overview
BridgeFin is a digital-first consumer and small-business lending platform that operates as an internal division of BRB. BridgeFin is not a separately incorporated legal entity, a variable interest entity, or a joint venture; rather, it functions as an operating division of the Bank, with its assets, liabilities, and operations consolidated within BRB's financial statements. The BridgeFin division was launched in the third quarter of 2020 (Q3 2020) as part of BRB's strategic initiative to expand its presence in the digital lending market and to diversify its loan portfolio beyond traditional commercial and residential lending products.
The BridgeFin division is led by Naveen Chakravarti, Senior Vice President, who reports directly to Patricia Navarrete, CFO. Mr. Chakravarti oversees all aspects of BridgeFin's operations, including product development, origination, servicing, and credit-risk management within the division.
BridgeFin offers two primary loan products:
1. Unsecured Personal Loans. Consumer personal loans originated through BridgeFin's online and mobile application platforms, with an average loan size of approximately $18,400. These loans are typically offered with fixed interest rates and terms ranging from 24 to 60 months. Borrower eligibility and pricing are determined through a proprietary AI-driven underwriting model (the "BridgeFin Underwriting Engine") that evaluates traditional credit bureau data, alternative data sources, and behavioral analytics.
2. Small-Business Microloans. Unsecured and lightly secured microloans extended to small businesses and sole proprietors, with an average loan size of approximately $42,000 and terms generally ranging from 12 to 36 months. These loans are also originated through digital channels and evaluated by the BridgeFin Underwriting Engine, with additional business-specific financial data incorporated into the credit decision.
BridgeFin's FY 2022 origination volume was approximately $2.87 billion across both product lines, reflecting rapid growth since the division's inception. As of December 31, 2022, the total outstanding loan balance in the BridgeFin Portfolio was approximately $4.1 billion, representing approximately 10.7% of BRB's total consolidated assets. The growth trajectory of the BridgeFin Portfolio — from inception in Q3 2020 to $4.1 billion in outstanding balances within approximately two and a half years — is relevant context for the CECL analysis, as the relative immaturity of the portfolio means that historical loss data is limited in duration and that vintage-cohort assumptions carry heightened significance in loss-rate modeling.
C. The SEC Investigation
The SEC's investigation of BRB's credit-loss accounting for the BridgeFin Portfolio has proceeded through the following key stages:
Voluntary Request Phase. On September 18, 2023, BRB received a voluntary request for documents and information from the SEC Division of Enforcement. The request was directed to BRB's General Counsel, Derek Sung, and focused on the Bank's CECL methodology, BridgeFin loan modification and re-aging practices, ACL calculations, and related internal communications for the period January 1, 2022 through June 30, 2023.
Formal Order of Investigation. On October 12, 2023, the SEC issued a Formal Order of Investigation, Order No. EN-2023-0947, authorizing the Staff to conduct a formal investigation into possible violations of the federal securities laws in connection with BRB's financial reporting for the BridgeFin Portfolio. The issuance of the Formal Order conferred upon the Staff subpoena authority with respect to documents and testimony.
Document Subpoena. On October 15, 2023, the Staff issued a document subpoena (designated HO-14287-001) requiring BRB to produce documents relating to, among other things: the BridgeFin CECL model and all supporting documentation; loan-level data for modified and re-aged loans; internal and external audit work papers; Audit Committee materials; and communications among senior management, the risk function, and the external auditors concerning BridgeFin credit quality and loss provisioning.
The SEC Staff members assigned to the investigation include the following individuals, as identified through correspondence and the Formal Order:
• Senior Staff Attorney: Andrea Felton
• Staff Attorney: Christopher Yuen
• Staff Accountant: Diane Milosevic
• Assistant Director, Financial Fraud Unit: Gregory Bonham
• Branch Chief: Renata Moss
The core issue under investigation, as articulated by the Staff in its correspondence and through the investigative testimony process, is whether BRB's Annual Report on Form 10-K for FY 2022 (filed February 28, 2023), its Quarterly Report on Form 10-Q for Q1 2023 (filed May 10, 2023), and its Quarterly Report on Form 10-Q for Q2 2023 (filed August 9, 2023) materially understated expected credit losses in the BridgeFin Portfolio. The Staff's inquiry centers on BRB's treatment of re-aged and modified loans within its vintage-cohort CECL model and whether the resulting ACL figures failed to incorporate deteriorating credit-quality metrics in a manner that rendered BRB's periodic reports inaccurate or misleading.
The potential statutory and regulatory provisions under consideration by the Staff include the following:
• Section 13(a) of the Securities Exchange Act of 1934 (15 U.S.C. § 78m(a)) — requiring the filing of accurate periodic reports with the Commission;
• Section 13(b)(2)(A) of the Securities Exchange Act of 1934 (15 U.S.C. § 78m(b)(2)(A)) — requiring that an issuer's books, records, and accounts, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the issuer;
• Rule 13a-1 (17 C.F.R. § 240.13a-1) — requiring the filing of annual reports; and
• Rule 12b-20 (17 C.F.R. § 240.12b-20) — requiring that reports contain such further material information as is necessary to make the required statements, in light of the circumstances under which they are made, not misleading.
We note that the provisions identified above — Sections 13(a) and 13(b)(2)(A) and the rules promulgated thereunder — are non-scienter provisions; that is, the SEC is not required to prove that BRB acted with the intent to deceive, manipulate, or defraud in order to establish a violation. Nonetheless, as discussed further in Section VIII below, the Staff's investigation also appears to have evaluated whether evidence of scienter exists that might support fraud-based charges under Section 10(b) of the Exchange Act and Rule 10b-5 promulgated thereunder. As set forth below, no evidence of such scienter has emerged.
D. Engagement of Calderwood & Harkness and Brindoval Analytic Partners
Calderwood & Harkness LLP. The law firm of Calderwood & Harkness LLP, with offices at One Manhattan West, 37th Floor, New York, New York 10001, was retained by BRB as defense counsel in connection with the SEC investigation effective September 20, 2023. The engagement is designated as Matter No. SC-MAT-2023-4172. The engagement team consists of:
• Responsible Partner: Victoria Chambers
• Billing Partner: Howard Quinn
• Lead Associate: Jasmine Thornton
• Working Associate: Liam Donahue
Brindoval Analytic Partners LLC. The accounting and consulting firm of Brindoval Analytic Partners LLC, with offices at 350 South Grand Avenue, Suite 2100, Los Angeles, California 90071, was retained by Calderwood & Harkness under the Kovel doctrine and the attorney work-product protection effective October 10, 2023. Brindoval's retention was effected through a consulting engagement agreement between Calderwood & Harkness and Brindoval, pursuant to which Brindoval's work is performed at the direction of and for the benefit of counsel, and all work product, analyses, and communications are protected by the attorney work-product doctrine. The Brindoval engagement is led by Dr. Franklin Osei, CPA, CFF, a forensic accounting specialist with more than twenty years of experience in financial institution regulatory matters, credit-loss modeling, and SEC enforcement defense. The purpose of the Brindoval engagement is to perform a forensic accounting analysis of BRB's ACL methodology and BridgeFin loss-rate modeling, and to provide expert support in connection with any Wells Submission or subsequent proceedings.
Independent Auditor. BRB's independent registered public accounting firm for FY 2021 and FY 2022 was Whitfield Keane LLP (PCAOB Registration No. 2847). The engagement partner for the FY 2022 audit was Lorraine Dupont. Whitfield Keane audited BRB's consolidated financial statements for FY 2022 and issued an unqualified audit opinion. Whitfield Keane is separately represented in connection with the SEC investigation by the law firm of Hargrove & Millstein LLP, with lead partner Samuel Hargrove. Calderwood & Harkness has coordinated with Hargrove & Millstein on matters of mutual relevance to the extent consistent with the preservation of privilege, but this memorandum reflects solely the analysis and conclusions of Calderwood & Harkness and Brindoval.
Securities Reporting Counsel. BRB's regular securities reporting counsel is the law firm of Caldwell Pratt & Stein LLP, with lead partner Martin Caldwell. Caldwell Pratt & Stein prepared and reviewed the periodic filings at issue in the investigation, including the FY 2022 Form 10-K and the Q1 and Q2 2023 Forms 10-Q. Caldwell Pratt & Stein is not a party to this memorandum and is separately represented in connection with any inquiries directed to their role in the preparation of the filings.
III. Framework: CECL (ASC 326) and Applicable Accounting Standards
A. Overview of ASC 326 — Financial Instruments: Credit Losses
The current expected credit losses model, codified at ASC 326, Financial Instruments — Credit Losses, was issued by the Financial Accounting Standards Board ("FASB") in June 2016 as Accounting Standards Update ("ASU") 2016-13. ASC 326 fundamentally altered the accounting for credit losses on financial instruments measured at amortized cost by replacing the prior "incurred loss" model — under which entities recognized credit losses only when it was probable that a loss had been incurred — with a forward-looking "expected loss" model. Under ASC 326, entities are required to estimate and recognize expected credit losses over the full remaining contractual life of financial assets at the reporting date, resulting in earlier recognition of credit losses and, in many cases, larger allowances.
BRB adopted ASC 326 effective January 1, 2020, which was the mandatory adoption date for public business entities that are SEC filers and that meet the definition of large accelerated filers. As noted above, BRB effected the adoption through a modified retrospective transition approach, recording a cumulative-effect adjustment to retained earnings as of the adoption date.
The following key principles of ASC 326 are relevant to the analysis in this memorandum:
Lifetime Expected Credit Losses. ASC 326-20-30-1 requires that the allowance for credit losses reflect management's estimate of all expected credit losses over the contractual term of the financial asset, adjusted for expected prepayments but not adjusted for expected extensions, renewals, or modifications unless the entity reasonably expects to execute a troubled debt restructuring with the borrower. This "life of loan" approach means that the ACL must capture not only losses that are currently evident but also losses that are expected to occur in future periods based on current conditions and reasonable forecasts.
Reasonable and Supportable Forecasts. ASC 326-20-30-7 through 30-9 require entities to consider available information relevant to assessing the collectibility of cash flows, including historical loss information, current conditions, and "reasonable and supportable forecasts." The standard acknowledges that an entity may not be able to develop reasonable and supportable forecasts over the entire contractual life of a financial asset and provides that, for periods beyond which the entity can make reasonable and supportable forecasts, the entity shall revert to historical loss information. This forecast-plus-reversion structure is a critical component of BRB's CECL model, as discussed in Section IV.
Range of Reasonable Estimates. The CECL framework explicitly contemplates that the measurement of expected credit losses involves significant judgment and that a range of acceptable estimates may exist. ASC 326-20-30-3 provides that the estimate of expected credit losses should reflect "an entity's current estimate of all expected credit losses," and the Basis for Conclusions to ASU 2016-13 (paragraphs BC64-BC67) emphasizes that the Board did not intend for the standard to require a single deterministic outcome. Rather, the ACL is intended to be the entity's best estimate within a range of possible outcomes, informed by the entity's assessment of the quality and relevance of available information. This principle is central to BRB's defense: the fact that a different set of reasonable assumptions could produce a different ACL figure does not, standing alone, render BRB's reported ACL unreasonable or its financial statements materially misstated.
Permitted Methodologies. ASC 326-20-30-2 does not prescribe a specific methodology for estimating expected credit losses. The standard permits entities to use any methodology that is consistent with the principles of the standard, including but not limited to: vintage analysis, probability-of-default/loss-given-default ("PD/LGD") models, discounted cash flow analysis, loss-rate methods, roll-rate methods, and combinations thereof. This methodological flexibility reflects the FASB's recognition that different portfolio characteristics and data environments may call for different approaches. BRB's selection of a vintage-cohort loss-rate model for the BridgeFin Portfolio, as described in Section IV, is squarely within the range of methodologies contemplated by the standard.
Qualitative Adjustments. ASC 326-20-30-9A and related guidance acknowledge that quantitative models may not fully capture all factors affecting expected credit losses. Entities are permitted — and, where relevant, expected — to incorporate qualitative adjustments ("Q-factors") to reflect conditions and trends not adequately captured by quantitative inputs. These qualitative adjustments must be directionally consistent with the conditions they are intended to address and should be supported by appropriate documentation. BRB's use of qualitative overlays in its BridgeFin CECL model is discussed in Section IV.D.
B. Vintage Cohort Analysis
BRB employs a vintage-cohort approach as its primary loss-estimation methodology for the BridgeFin Portfolio. Under this approach, loans are grouped by origination quarter (the "vintage"), and historical charge-off and loss rates are tracked by vintage over the life of each cohort. The observed loss patterns for matured or seasoned vintages are then used to project lifetime expected losses for less-seasoned vintages, with adjustments for current conditions and reasonable and supportable forecasts.
Vintage-cohort analysis is one of the most commonly used CECL methodologies for consumer loan portfolios, as recognized in the implementation guidance provided in ASC 326-20-55 and in the interagency guidance issued by the federal banking agencies (including the OCC) in connection with CECL adoption. The methodology is particularly well-suited for portfolios of homogeneous loans — such as the unsecured personal loans and small-business microloans originated through BridgeFin — because it allows the entity to observe and project loss emergence patterns at a granular level while maintaining the statistical power that comes from pooling large numbers of similar loans within each cohort.
Key features of BRB's vintage-cohort approach include the grouping of loans by origination quarter, the segmentation of cohorts by product type and internal credit-risk grade, and the application of historical loss rates adjusted for forecasted conditions. The specific architecture and inputs of BRB's model are described in greater detail in Section IV below.
C. Treatment of Loan Modifications and Re-Aging Under CECL
The treatment of modified and re-aged loans within a vintage-cohort CECL framework is the central methodological issue in this investigation. It is important to note at the outset that ASC 326 does not prescribe a single required treatment for modified or re-aged loans in the context of vintage-cohort analysis. The standard's principles-based approach affords entities discretion in determining how to classify and track modified loans, provided that the chosen approach is consistently applied, adequately supported, and appropriately disclosed.
In practice, two broadly accepted approaches have emerged among financial institutions implementing CECL through vintage-cohort models:
Approach 1 — Cohort Reset (BRB's Approach). Under this approach, loans that are modified (e.g., through payment deferrals, term extensions, or interest-rate reductions) are treated as "newly performing" from the modification date. The loan's delinquency status is reset to "current," and the loan is migrated to a new or adjusted cohort reflecting post-modification performance expectations. The rationale for this approach is that the modification itself fundamentally changes the risk profile and expected cash flows of the loan. A borrower who has received a payment deferral, for example, may have a materially different probability of default going forward than a borrower in the same vintage who has not been modified. By placing the modified loan into a new or adjusted cohort, the model captures the post-modification loss trajectory, which is presumed to differ from the pre-modification trajectory. This approach is particularly relevant when post-modification performance data demonstrates that modified loans perform materially better (or differently) than unmodified delinquent loans.
Approach 2 — Vintage Retention (SEC Staff's Approach). Under this approach, modified loans are retained in their original vintage cohort, and the original delinquency history is preserved for loss-rate calculation purposes. The rationale for this approach is that the pre-modification delinquency history remains predictive of future loss and that re-aging may mask underlying credit deterioration. Proponents of this approach argue that a loan that was 90 days delinquent before modification still carries the credit risk associated with that delinquency, even if the borrower has technically been brought current through a modification arrangement.
Neither approach is prohibited under ASC 326. The choice between them is a matter of management judgment, subject to the overarching requirements that the chosen methodology be (i) applied consistently across the portfolio, (ii) supportable based on relevant data and analysis, and (iii) accompanied by adequate disclosure that enables investors to understand the methodology and its impact on the reported ACL. As discussed in Section IV.C, BRB chose Approach 1 (Cohort Reset), documented its rationale, and applied the methodology consistently across all modified BridgeFin loans throughout FY 2022. The SEC Staff's preliminary view rests on the application of Approach 2 (Vintage Retention), which produces a higher ACL estimate but does not represent the sole permissible approach.
We note that ASU 2022-02, Financial Instruments — Credit Losses (Topic 326): Troubled Debt Restructurings and Vintage Disclosures, which was effective for BRB beginning January 1, 2023, eliminated the accounting concept of "troubled debt restructurings" and expanded vintage disclosure requirements. While ASU 2022-02 postdates the FY 2022 reporting period at issue, BRB's voluntary adoption of enhanced vintage disclosures in Q3 2023 (discussed in Section VII.C) reflects responsiveness to the evolving disclosure framework.
D. Regulatory Guidance
In addition to ASC 326, BRB's ACL methodology is informed by supervisory guidance from the OCC and the federal banking agencies. The OCC's Handbook on Allowance for Credit Losses (revised April 2023) provides comprehensive guidance to national banks on the implementation and maintenance of ACL methodologies under the CECL framework. Key principles emphasized in the OCC Handbook include:
• Institutions must use methodologies appropriate to the nature, size, and complexity of their loan portfolios;
• The ACL should reflect management's best estimate of expected credit losses, considering all available relevant information;
• Reasonable management judgments are expected and are a necessary part of the credit-loss estimation process;
• Institutions should document their methodologies, assumptions, and the basis for significant judgments;
• Internal controls over the ACL estimation process should be adequate to ensure the reliability and integrity of the reported figures.
It is significant that BRB's most recent OCC examination, completed as of September 2023, resulted in a composite CAMELS rating of 2 (Satisfactory), with no material deficiency cited in the Bank's ACL methodology. While the OCC examination and the SEC investigation are separate regulatory processes with distinct standards, the OCC's Satisfactory rating provides meaningful corroborative evidence that BRB's credit-loss estimation practices — including the treatment of modified loans — were within the range of acceptable practices recognized by the Bank's primary prudential regulator.
IV. BRB's CECL Model — Structure and Key Inputs
A. Model Architecture
BRB's CECL model for the BridgeFin Portfolio is a vintage-cohort loss-rate model that incorporates three primary segmentation dimensions:
1. Product Type. Loans are first segmented by product type: unsecured consumer personal loans and small-business microloans. This segmentation reflects the distinct risk characteristics, borrower profiles, and loss behaviors of the two product lines.
2. Origination Quarter (Vintage). Within each product type, loans are grouped into vintage cohorts based on the quarter in which they were originated. Each vintage cohort tracks the performance of all loans originated during a given quarter, allowing the model to observe and compare loss emergence patterns across cohorts and over time.
3. Internal Credit-Risk Grade. Within each vintage cohort, loans are further stratified by the internal credit-risk grade assigned at origination by the BridgeFin Underwriting Engine. The AI-driven underwriting model assigns each loan a numeric risk score (on a scale of 1 to 10, with 1 being the highest credit quality and 10 being the lowest), which is mapped to one of five risk-grade categories: Prime (scores 1–2), Near-Prime (3–4), Non-Prime (5–6), Sub-Prime (7–8), and Deep Sub-Prime (9–10). This risk-grade stratification allows the model to differentiate loss expectations within each vintage based on the initial credit quality of the loans.
Loss rates are estimated based on historical charge-off experience by vintage, with adjustments made to reflect current conditions and reasonable and supportable forecasts. The model's forecast horizon extends eight quarters beyond the reporting date (i.e., for the FY 2022 year-end calculation, the forecast period ran from January 1, 2023 through December 31, 2024). Beyond the eight-quarter forecast horizon, the model reverts to long-run historical averages, calculated using available BridgeFin performance data supplemented by proxy data from comparable consumer lending portfolios where BridgeFin's limited operating history (approximately 2.5 years as of the FY 2022 reporting date) was insufficient to develop robust long-run loss estimates. The reversion methodology employs a straight-line mean reversion over four additional quarters (the "reversion period"), transitioning from the forecasted loss rates to the long-run historical average over a one-year glide path.
The overall model architecture, including the choice of vintage-cohort methodology, the segmentation dimensions, the forecast horizon, the reversion methodology, and the calibration approach, was developed by BRB's quantitative risk analytics team under the supervision of Annette Kowalski, Chief Risk Officer, with input from external model validation consultants. The model was initially implemented in connection with BRB's adoption of ASC 326 in 2020 and was updated in Q3 2021 to incorporate BridgeFin-specific performance data as the portfolio reached sufficient size and seasoning to support disaggregated analysis.
B. Key Quantitative Inputs (FY 2022 Year-End)
The following table summarizes the key quantitative inputs and outputs of BRB's CECL model for the BridgeFin Portfolio as of December 31, 2022:
These inputs were derived from BRB's loan-level data systems and were subjected to reconciliation procedures as part of the quarterly ACL estimation process. The model's output — the $143.5 million ACL — was reviewed and approved through the following governance chain:
• Initial Calculation and Review: Prepared by BRB's quantitative risk analytics team and reviewed by the BridgeFin finance team;
• Chief Accounting Officer Approval: Raymond Hu reviewed the model output, supporting documentation, and qualitative overlay justification, and approved the ACL figure for inclusion in the financial statements;
• Chief Risk Officer Oversight: Annette Kowalski reviewed the model's credit-risk assumptions and confirmed their consistency with observed portfolio trends and macroeconomic conditions;
• Audit Committee Review: The BRB Audit Committee, chaired by Gwendolyn Prescott, received a presentation on the BridgeFin ACL methodology and year-end results at its February 2023 meeting, prior to the filing of the FY 2022 Form 10-K;
• External Audit: Whitfield Keane LLP (Lorraine Dupont, engagement partner) audited the ACL as part of its audit of BRB's FY 2022 consolidated financial statements. Whitfield Keane evaluated the reasonableness of BRB's CECL methodology, tested the model inputs and calculations, assessed the qualitative overlay adjustments, and concluded that the reported ACL was within the range of reasonable estimates. Whitfield Keane issued an unqualified audit opinion on BRB's FY 2022 financial statements.
C. Re-Aging / Modification Practices
The treatment of loan modifications and re-aging within BRB's CECL model is the principal area of disagreement between BRB and the SEC Staff. The following provides a detailed description of BRB's practices.
During FY 2022, BRB modified a total of 12,847 BridgeFin loans with an aggregate outstanding principal balance of approximately $236.4 million. These modifications were granted pursuant to BRB's BridgeFin Loan Modification Policy (the current version of which, effective as of December 31, 2022, is referenced in Appendix 4) and were extended to borrowers experiencing temporary financial hardship. The modifications took three principal forms, in many cases applied in combination:
• Payment Deferrals: Deferral of one or more scheduled monthly payments to the end of the loan term, with the deferred amount added to the remaining principal balance or extended as additional contractual payments. Approximately 7,214 loans ($131.8 million) received payment deferrals.
• Term Extensions: Extension of the loan's maturity date by three to twelve months, reducing the required monthly payment while maintaining the original interest rate. Approximately 4,318 loans ($78.2 million) received term extensions.
• Interest-Rate Reductions: Temporary or permanent reduction in the stated interest rate to reduce the borrower's monthly payment obligation. Approximately 1,315 loans ($26.4 million) received interest-rate reductions. (Certain loans received more than one form of modification; the figures above reflect the primary modification type for each loan.)
Upon the execution of a modification, BRB's CECL model reclassified the modified loan as "current," resetting its delinquency clock to zero. The loan was then assigned to a new performance cohort — designated internally as a "Modified-Current" ("MC") cohort — for purposes of vintage-cohort loss-rate tracking. The MC cohort was further segmented by the quarter in which the modification occurred and by the original risk grade of the loan, allowing BRB to track post-modification performance patterns at a granular level.
The practical effect of this treatment was to remove the 12,847 modified loans from the delinquent or high-risk vintage buckets in which they had previously resided. Because these loans were disproportionately drawn from earlier vintages with higher observed delinquency rates, the removal reduced the weighted-average delinquency rate of those original vintages and, consequently, reduced the estimated loss rate applied to the remaining loans in those vintages. At the same time, the modified loans were assigned loss-rate expectations based on the observed performance of the MC cohort, which reflected a lower initial delinquency rate (by definition, all loans entered the MC cohort at "current" status) and, over time, a lower loss trajectory than the original delinquent vintages from which they were drawn.
BRB's rationale for this treatment, as documented in the Bank's CECL model governance documentation and in the working papers produced in the document production (Bates Nos. BRB-SEC-004217 through BRB-SEC-004489), is that the modification itself constitutes a credit event that fundamentally alters the risk profile and expected cash flows of the loan. Specifically, BRB documented the following bases for its re-aging methodology:
(a) Modified loans, by receiving payment relief, have a different go-forward payment schedule and a different expected cash-flow stream than unmodified loans in the same vintage;
(b) Post-modification performance data, while limited given the relative newness of the BridgeFin Portfolio, indicated that modified loans exhibited materially different loss trajectories than unmodified delinquent loans;
(c) Retaining modified loans in their original vintage cohorts, while treating them as "current," would distort the vintage-level loss rates by mixing two distinct performance populations within the same cohort; and
(d) Segregating modified loans into a separate MC cohort allowed BRB to track and model the unique credit-risk characteristics of the modified population more accurately.
This documentation, prepared contemporaneously with the FY 2022 ACL estimation process, supports the conclusion that BRB's re-aging treatment was a deliberate, documented methodological choice, not an ad hoc or reactive effort to suppress loss provisions.
D. Qualitative Overlay Adjustments
In addition to the quantitative output of the vintage-cohort loss-rate model, BRB applied qualitative overlay adjustments ("Q-factors") to the BridgeFin ACL estimation for FY 2022. The qualitative overlay process is a standard component of CECL implementation, reflecting the FASB's recognition that quantitative models may not fully capture all factors affecting expected credit losses. BRB's qualitative overlay process was governed by a written policy (the "Qualitative Factor Adjustment Policy," effective March 2021, revised January 2022) and was subject to review and approval by the Chief Accounting Officer and the Chief Risk Officer.
For the FY 2022 year-end ACL calculation, BRB applied qualitative adjustments for the following factors:
• Macroeconomic Conditions. BRB considered prevailing and forecasted macroeconomic conditions, including labor market trends (the national unemployment rate, which was 3.5% as of December 2022), consumer spending patterns, inflation data, and Federal Reserve monetary policy actions. BRB's macroeconomic overlay reflected a moderately adverse scenario, incorporating expectations for a gradual increase in unemployment and a deceleration in consumer spending growth over the eight-quarter forecast horizon.
• Changes in Underwriting Criteria. BRB tightened certain BridgeFin underwriting criteria in Q3 2022, including raising minimum credit-score thresholds and reducing maximum loan-to-income ratios for the personal loan product. The qualitative overlay incorporated an adjustment to reflect the expected improvement in credit quality for loans originated after the underwriting changes, partially offset by the anticipated seasoning of pre-tightening vintages.
• Portfolio Growth and Seasoning Effects. Given the rapid growth of the BridgeFin Portfolio and its relatively short operating history, BRB applied a qualitative adjustment to account for the increased uncertainty associated with projecting lifetime losses for a portfolio that had not yet experienced a full economic cycle. This adjustment recognized that the historical loss data available for calibrating the vintage-cohort model was limited in duration and may not fully capture the loss experience that would emerge under stressed economic conditions.
The total qualitative overlay added approximately $12.8 million to the quantitative model output. This figure was derived from a structured assessment process in which each Q-factor was evaluated on a five-point severity scale (minimal, low, moderate, elevated, high) and assigned a corresponding basis-point adjustment to the base loss rate. The qualitative overlay was documented in a memorandum prepared by BRB's quantitative risk analytics team and reviewed by Raymond Hu and Annette Kowalski (Bates Nos. BRB-SEC-005102 through BRB-SEC-005178).
It is worth noting that the Staff's preliminary ACL figure of $180.8 million was calculated using the Staff's alternative vintage-retention methodology combined with the Staff's own qualitative assumptions, which were more conservative than BRB's in certain respects. As discussed in Section V.C, Brindoval's sensitivity analysis demonstrates that the difference between the two approaches narrows substantially when the vintage-retention methodology is combined with BRB's qualitative overlay calibration, indicating that the qualitative assumptions are a significant driver of the gap between the two estimates.
V. SEC Staff's Preliminary View — Analysis and Response
A. Staff's Contention
Through investigative correspondence dated November 14, 2023 and supplemented by Staff Accountant Diane Milosevic's written analysis provided to Calderwood & Harkness on November 28, 2023 (the "Staff Analysis"), the SEC Staff has communicated its preliminary view that BRB's ACL for the BridgeFin Portfolio should have been approximately $180.8 million as of December 31, 2022. This figure represents a coverage ratio of approximately 4.41% of the $4.1 billion in outstanding BridgeFin loans, compared to BRB's reported ACL of $143.5 million (3.50%). The potential understatement identified by the Staff is therefore approximately $37.3 million.
The Staff Analysis rests primarily on the following contentions:
1. Vintage Retention for Modified Loans. The Staff contends that the 12,847 modified BridgeFin loans ($236.4 million in aggregate outstanding balance) should have been retained in their original vintage delinquency buckets for loss-rate estimation purposes, rather than reclassified to "current" status and migrated to a new MC cohort. The Staff's position is that the pre-modification delinquency history of these loans remains the best predictor of future loss and that re-aging obscures the true credit-risk profile of the BridgeFin Portfolio.
2. Recalculated Delinquency and Charge-Off Metrics. Under the Staff's vintage-retention approach, the BridgeFin Portfolio's key credit-quality metrics would be recalculated as follows:
3. Alleged Masking of Credit Deterioration. The Staff further contends that BRB's re-aging practice had the effect of masking deteriorating credit-quality trends in the BridgeFin Portfolio, particularly in the earlier vintages (Q3 2020 through Q4 2021) that had experienced the highest delinquency rates. By removing the most troubled loans from these vintages and reclassifying them as "current," BRB's model allegedly produced artificially favorable vintage-level loss rates that understated expected losses.
4. Disclosure Deficiency. The Staff has also raised concerns that BRB's FY 2022 Form 10-K and Q1-Q2 2023 Forms 10-Q did not adequately disclose the magnitude of the re-aging practice or its impact on reported credit-quality metrics, thereby depriving investors of material information necessary to evaluate the BridgeFin Portfolio's credit risk.
We address each of these contentions below and in the sections that follow.
B. Materiality Analysis
The threshold question in evaluating the Staff's preliminary view is whether the $37.3 million difference between BRB's reported ACL and the Staff's alternative figure constitutes a "material" misstatement within the meaning of the federal securities laws and applicable accounting standards.
Quantitative Analysis. BRB's internal materiality threshold, established for purposes of its annual financial statement preparation and audit process, is calculated as 5% of pre-tax income. For FY 2022, this threshold is:
$612 million × 5% = $30.6 million
The Staff's alleged understatement of $37.3 million exceeds this threshold by approximately $6.7 million (21.9% above the threshold). On a purely arithmetic basis, therefore, the difference is above BRB's internally applied quantitative materiality benchmark.
However, materiality under the federal securities laws is not determined by rigid quantitative tests alone. The Supreme Court has defined information as "material" if there is a "substantial likelihood that the disclosure of the omitted fact would have been viewed by the reasonable investor as having significantly altered the 'total mix' of information made available." TSC Industries, Inc. v. Northway, Inc., 426 U.S. 438, 449 (1976); Basic Inc. v. Levinson, 485 U.S. 224, 231-32 (1988). The SEC's own guidance in Staff Accounting Bulletin No. 99 ("SAB 99") further provides that quantitative materiality thresholds are merely "initial steps" in assessing materiality, and that the assessment must also consider qualitative factors. SAB 99 identifies a non-exclusive list of qualitative factors that may render a quantitatively small misstatement material, including whether the misstatement:
• Masks a change in earnings or other trends;
• Hides a failure to meet analysts' consensus expectations;
• Changes a loss into income or vice versa;
• Concerns a segment or other portion of the registrant's business that has been identified as playing a significant role in the registrant's operations or profitability;
• Affects the registrant's compliance with regulatory requirements;
• Affects the registrant's compliance with loan covenants or other contractual requirements;
• Has the effect of increasing management's compensation; or
• Involves concealment of an unlawful transaction.
Qualitative Analysis — Brindoval's Assessment. Dr. Osei and the Brindoval team conducted a comprehensive qualitative materiality assessment applying the SAB 99 factors. The key findings of this assessment are as follows:
Earnings Trend. The $37.3 million difference, if recognized as additional provision expense in FY 2022, would have reduced BRB's pre-tax income from $612 million to approximately $574.7 million — a reduction of approximately 6.1%. While not insignificant, this reduction would not have altered the directional trend of BRB's earnings: BRB's pre-tax income grew from $547 million in FY 2021 to $612 million in FY 2022, and the adjusted figure of $574.7 million would still represent year-over-year growth of approximately 5.1%. The additional provision would not have masked a decline in earnings or reversed an earnings trend.
Analyst Expectations. BRB's reported diluted EPS for FY 2022 was $6.14. The consensus analyst estimate (per Vireo Software data as of February 27, 2023) was $5.98. If the additional $37.3 million provision (approximately $28.7 million after tax, assuming a 23% effective rate) were recognized, diluted EPS would have been approximately $5.77. This figure would have been below the consensus estimate by approximately $0.21 per share, or approximately 3.5%. Brindoval notes, however, that the consensus range of analyst estimates extended from $5.42 to $6.38, and the adjusted EPS of $5.77 would have fallen within this range. Moreover, the question under SAB 99 is whether the misstatement "hides a failure to meet analysts' consensus expectations," which contemplates situations where management intentionally manages earnings to meet or beat consensus. There is no evidence in the record that BRB's ACL methodology was calibrated to achieve a specific EPS target, and the adjustment would not have converted a "miss" to a "beat" — rather, BRB's reported results exceeded consensus by $0.16, and the adjustment would have shifted the result to a miss of $0.21. This factor, in context, weighs modestly against a finding of materiality, as the consensus-beat margin was relatively narrow and was not the product of a last-minute or unusual adjustment.
Loss to Income Conversion. The adjustment does not convert income to a loss or vice versa. BRB would have reported substantial positive net income ($442.5 million after tax on an adjusted basis) even with the additional provision.
Segment Significance. The BridgeFin Portfolio represents approximately 10.7% of BRB's total assets and generated approximately 14% of total interest income in FY 2022. While BridgeFin is a significant and growing component of BRB's business, it is not the dominant segment, and the ACL for BridgeFin was separately disclosed as a line item in BRB's financial statements.
Regulatory Compliance. BRB's capital ratios remained well above all applicable regulatory minimums both as reported and on an adjusted basis. The additional $37.3 million provision would have reduced BRB's Common Equity Tier 1 ("CET1") capital ratio by approximately 7 basis points (from 11.48% to approximately 11.41%), well above the 6.5% well-capitalized threshold (including the capital conservation buffer).
Compensation Impact. There is no evidence that the ACL figure directly affected management compensation calculations. BRB's executive compensation program, as disclosed in its FY 2022 proxy statement, ties annual incentive compensation to pre-provision, pre-tax net revenue and return on equity, not to post-provision earnings.
Concealment of Unlawful Activity. There is no evidence or allegation that BRB's re-aging practice was used to conceal unlawful transactions.
Conclusion on Materiality. Brindoval's analysis concludes that the $37.3 million difference, while quantitatively exceeding BRB's 5% internal materiality threshold by $6.7 million, does not trigger the qualitative indicators of SAB 99 materiality in a manner that, taken together, would compel a conclusion of material misstatement. The difference arises from a discrete methodological judgment, not from a pervasive error or intentional manipulation, and does not alter the overall picture of BRB's financial condition or results of operations that was presented to investors. We acknowledge that this is a close case on quantitative grounds, and that the SEC Staff may weigh the analyst consensus factor differently than Brindoval has done. However, the balance of the qualitative factors supports the conclusion that the difference is not material in the context of the total mix of information available to investors.
C. Brindoval's Counter-Analysis
Dr. Osei and the Brindoval team performed an independent reconstruction of BRB's CECL model for the BridgeFin Portfolio, using loan-level data obtained from BRB's document production (Bates Nos. BRB-SEC-012001 through BRB-SEC-018947) and the Bank's CECL model documentation (Bates Nos. BRB-SEC-004217 through BRB-SEC-005178). Brindoval's analysis was conducted using three methodological approaches:
(a) BRB's Approach (Cohort Reset): Replication of BRB's reported methodology, resetting modified loans to "current" status and migrating them to the MC cohort;
(b) Staff's Approach (Vintage Retention): Application of the Staff's alternative methodology, retaining modified loans in their original vintage delinquency buckets and preserving pre-modification delinquency history;
(c) Hybrid Approach: A blended methodology in which modified loans are retained in their original vintage cohorts but assigned an adjusted (lower) loss rate reflecting the post-modification concession, rather than the full loss rate applicable to the unmodified delinquent population.
Brindoval's key findings are as follows:
Finding 1: Range of Reasonable Estimates
When reasonable variations in key assumptions — including the forecast horizon (6, 8, or 10 quarters), the reversion methodology (straight-line vs. weighted-average mean reversion), the proxy data selection for long-run historical loss rates, and the qualitative overlay calibration — are applied across all three methodological approaches, the resulting ACL for the BridgeFin Portfolio ranges from approximately $138 million to $185 million. This range represents the "zone of reasonable estimation" within which informed management judgment may properly place the ACL.
BRB's reported ACL of $143.5 million falls within this range, positioned in the lower half but clearly within the bounds of reasonable estimation. The Staff's alternative ACL of $180.8 million also falls within this range, positioned in the upper portion. Both estimates represent permissible points within the spectrum of outcomes produced by reasonable methodological choices. The fact that the Staff's estimate is higher than BRB's does not, in itself, demonstrate that BRB's estimate was unreasonable; it demonstrates that a legitimate methodological disagreement exists.
The following table illustrates the sensitivity of the ACL to key assumptions under each of the three approaches:
Finding 2: Re-Aging Treatment Is Defensible
BRB's practice of resetting modified loans to "current" status and assigning them to a new performance cohort is a recognized methodology in the banking industry. Brindoval's review of publicly available CECL disclosures, regulatory guidance, and industry practice materials confirms that this approach is used by other banking institutions of comparable size and portfolio composition and is not prohibited by ASC 326 or any authoritative interpretive guidance.
The re-aging treatment is analytically sound when post-modification performance data indicates that modified loans exhibit a materially different loss trajectory than unmodified delinquent loans. As discussed in Finding 3 below, BRB's post-modification performance data supports this conclusion.
Finding 3: Post-Modification Performance
Brindoval's analysis of loan-level performance data for the 12,847 modified BridgeFin loans as of December 31, 2022 reveals the following:
• 78% (10,021 loans) remained current or were fewer than 30 days past due as of December 31, 2022;
• 13% (1,670 loans) were 30–89 days past due;
• 9% (1,156 loans) were 90 or more days past due or had been charged off subsequent to modification.
The 78% current-or-near-current rate among modified loans provides meaningful support for BRB's judgment that modification fundamentally altered the credit-risk profile of the affected loans. By contrast, the comparable population of unmodified loans in the same original vintage cohorts that were 30+ days delinquent at any point during FY 2022 exhibited a re-default rate of approximately 41% within six months, roughly 4.6 times the 9% severe-delinquency/charge-off rate among modified loans over a comparable observation period.
Brindoval acknowledges that the post-modification observation period is relatively short — the majority of FY 2022 modifications occurred in the second half of the year, meaning that only three to six months of post-modification performance data was available as of the December 31, 2022 reporting date. This limited observation period introduces uncertainty into the analysis, and it is possible that the re-default rate among modified loans will increase as the observation period lengthens. However, the available data as of the reporting date supported BRB's judgment, and ASC 326 requires management to use the information available at the reporting date, not to apply hindsight.
Finding 4: Vintage Analysis Sensitivity
Brindoval ran detailed sensitivity analyses to isolate the drivers of the difference between BRB's reported ACL and the Staff's alternative figure. The results demonstrate that the $37.3 million difference is not attributable solely to the vintage-cohort treatment of modified loans but is also driven by differences in qualitative overlay assumptions.
When the Staff's vintage-retention methodology is applied but combined with BRB's qualitative overlay calibration (rather than the Staff's more conservative qualitative assumptions), the resulting ACL is approximately $162.1 million — a difference of only approximately $18.6 million from BRB's reported figure of $143.5 million. This $18.6 million difference is below BRB's internal materiality threshold of $30.6 million and represents a spread of only 0.45 percentage points in coverage ratio (from 3.50% to 3.95%).
Conversely, the full $37.3 million difference ($180.8 million vs. $143.5 million) emerges only when the Staff's vintage-retention methodology is combined with the Staff's more conservative qualitative assumptions regarding macroeconomic forecasts and portfolio-seasoning adjustments. This finding demonstrates that approximately half of the identified difference ($18.7 million of $37.3 million) is attributable to differences in qualitative judgment, not to the vintage-cohort treatment of modified loans that is the central focus of the Staff's contention.
This decomposition is significant because qualitative overlay calibration is, by its nature, an exercise in judgment where a wide range of reasonable outcomes exists. The Staff's qualitative assumptions, while not unreasonable, are more conservative than BRB's in several respects — particularly with respect to the macroeconomic scenario weighting and the portfolio-seasoning adjustment — and these differences amplify the impact of the vintage-retention methodology.
VI. Peer Bank Comparison
A. Methodology
To provide additional context for the reasonableness of BRB's ACL methodology and reported coverage ratio, Brindoval identified and analyzed a peer group of twelve financial institutions with significant digital or fintech-originated consumer lending portfolios and comparable CECL disclosures. Peer institutions were selected based on the following criteria: (i) publicly traded bank or bank holding company; (ii) material consumer lending portfolio that includes digital or online-originated loans; (iii) disclosed CECL methodology for consumer lending segment; and (iv) comparable total asset size (within a range of approximately $15 billion to $75 billion in total assets).
Comparison metrics were drawn from publicly available sources, including annual reports on Form 10-K, quarterly reports on Form 10-Q, and regulato