post-testimony-debrief-hu.docx

1045-00007 · Client 1045 · Witness Preparation

post-testimony-debrief-hu.docx DOCUMENT Litigation

31 / 35 ← → ✕

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT PREPARED IN ANTICIPATION OF LITIGATION NOT TO BE DISCLOSED WITHOUT PRIOR AUTHORIZATION

CALDERWOOD & HARKNESS LLP 425 Lexington Avenue, 30th Floor New York, New York 10017

February 15, 2024

MEMORANDUM

TO: Victoria Chambers, Partner Howard Quinn, Partner

FROM: Jasmine Thornton, Lead Associate

CC: Liam Donahue, Working Associate

RE: Post-Testimony Debrief — Investigative Testimony of Raymond Hu, Chief Accounting Officer, Belmont Ridge Bank NA (February 14, 2024)

MATTER: SEC Investigation File No. HO-14287 C&H Engagement No. SC-MAT-2023-4172

PRIVILEGE LEGEND: This memorandum is a privileged and confidential attorney work product of Calderwood & Harkness LLP. It is protected in its entirety by the attorney-client privilege and the attorney work product doctrine. It was prepared in anticipation of litigation and reflects the mental impressions, conclusions, opinions, and legal theories of counsel. This memorandum is not to be disclosed, distributed, forwarded, copied, or otherwise shared outside the Calderwood & Harkness LLP engagement team and authorized Belmont Ridge Bank NA personnel without the prior written approval of Victoria Chambers.

I. Executive Summary

Raymond Hu, Chief Accounting Officer of Belmont Ridge Bank National Association ("BRB" or the "Bank"), testified under oath on February 14, 2024, before staff of the United States Securities and Exchange Commission ("SEC") Division of Enforcement, pursuant to the Formal Order of Investigation No. EN-2023-0947 (the "Formal Order"). The testimony was conducted at the SEC Division of Enforcement offices located at 100 F Street NE, Washington, D.C. 20549, and lasted approximately five and a half hours, inclusive of a mid-morning break and a lunch recess.

Hu's testimony was the third and final scheduled investigative testimony in this matter. The prior two witnesses testified as follows: (i) Patricia Navarrete, Chief Financial Officer of BRB, testified on January 18, 2024; and (ii) Naveen Chakravarti, Head of BRB's BridgeFin Division, testified on January 30, 2024. With the completion of Hu's testimony, the Staff has now examined the three principal BRB officers most directly involved in the matters under investigation — namely, BRB's accounting for the allowance for credit losses ("ACL") on the BridgeFin loan portfolio under ASC 326 (the Current Expected Credit Losses, or "CECL," standard) and related financial reporting disclosures.

Hu was represented at the testimony by Victoria Chambers (Partner) and Jasmine Thornton (Lead Associate) of Calderwood & Harkness LLP. Liam Donahue (Working Associate) attended in a supporting capacity for note-taking and document management. SEC Staff present at the testimony included Senior Staff Attorney Andrea Felton, who served as the lead questioner for approximately the first two hours; Staff Attorney Christopher Yuen, who asked a discrete set of questions concerning internal controls over financial reporting; and Staff Accountant Diane Milosevic, who served as the lead questioner for approximately three hours and focused on the technical mechanics of BRB's CECL model, with particular emphasis on the treatment of re-aged and modified BridgeFin loans.

Overall Assessment: Hu performed well throughout the testimony and presented as credible, methodical, and precise. His extensive background in public accounting — including fourteen years at Whitfield Keane LLP, BRB's independent auditor — was evident in his facility with accounting standards, audit terminology, and the granular mechanics of CECL model construction. He maintained composure and consistency even under sustained and technically detailed questioning from Milosevic, whose line of inquiry was notably more granular and technically sophisticated than we had anticipated. Milosevic's questioning suggests that the Staff has developed a deep and well-informed understanding of BRB's CECL model mechanics, likely aided by an outside accounting consultant or by Milosevic's own considerable expertise. Notwithstanding the rigor of the Staff's questioning, Hu's testimony effectively conveyed that BRB's CECL methodology was deliberate, well-documented, consistent with applicable accounting standards, and within the range of reasonable estimates under ASC 326. No significant inconsistencies with the prior testimonies of Navarrete or Chakravarti were identified. The testimony reinforces the core defense position for the anticipated Wells Submission.

II. Witness Background and Preparation

A. Professional Background

Raymond Hu has served as Chief Accounting Officer of BRB since June 2019. Prior to joining BRB, Hu spent fourteen years at Whitfield Keane LLP (PCAOB Registration No. 2847), BRB's independent auditor, where he held positions of increasing responsibility within the firm's audit practice. During his tenure at Whitfield Keane, Hu rose to the level of Senior Manager and was involved in the audits of several large regional and community banking institutions, although he was not a member of the BRB engagement team during his time at the firm. His decision to transition to the client side was contemporaneous with BRB's strategic expansion into digital lending through the BridgeFin platform, and he was hired in part for his deep expertise in credit-loss accounting and financial institution regulatory reporting.

At BRB, Hu reports directly to CFO Patricia Navarrete. He also maintains a dotted-line reporting relationship to Gwendolyn Prescott, Chair of BRB's Audit Committee. In his capacity as CAO, Hu was the primary BRB officer responsible for the design, implementation, and ongoing operation of the Bank's CECL model following BRB's adoption of ASC 326, which became effective for BRB on January 1, 2020 (its first fiscal year beginning after December 15, 2019, as required for SEC filers). He also supervised the preparation of the credit-loss disclosures appearing in BRB's periodic SEC filings, including the Management's Discussion and Analysis ("MD&A") and notes to the financial statements addressing the ACL.

B. Preparation Sessions

Witness preparation sessions were conducted on February 9, 11, and 13, 2024, at Calderwood & Harkness's New York offices. The sessions totaled approximately fourteen hours across the three days. The preparation team consisted of Victoria Chambers, Jasmine Thornton, and Liam Donahue.

The substance of the preparation sessions covered the following principal areas:

1. CECL model architecture and documentation — A comprehensive review of BRB's CECL model, including the probability-of-default / loss-given-default ("PD/LGD") framework, vintage-cohort segmentation methodology, and the specific parameters and assumptions applied to the BridgeFin portfolio.

2. Vintage cohort treatment and re-aging/modification practices — Detailed discussion of BRB's accounting treatment for modified and re-aged BridgeFin loans, including the reassignment of modified loans to new vintage cohorts, the rationale for this treatment, and the documentary support in BRB's model governance materials.

3. Quarterly true-up procedures and qualitative overlay adjustments — Review of BRB's process for quarterly recalibration of the CECL model, including the role of the Credit Loss Estimation Committee, the qualitative overlay factors applied, and the magnitude of overlay adjustments in the relevant periods.

4. Communications with Whitfield Keane LLP — Discussion of the cadence and substance of Hu's interactions with Whitfield Keane LLP engagement partner Lorraine Dupont and her audit team, particularly with respect to the FY 2022 year-end audit and the interim reviews for Q1 and Q2 2023.

5. Anticipated SEC focus areas — Identification of likely Staff focus areas based on the questioning observed during the Navarrete testimony (January 18, 2024) and the Chakravarti testimony (January 30, 2024), including the re-aging issue, the limited historical loss data for BridgeFin, the macroeconomic scenario weightings, and the potential materiality of the ACL difference.

6. Document familiarity — Review of key documents from BRB's production that we anticipated the Staff would use as exhibits during the testimony, including model governance memoranda, Credit Loss Estimation Committee meeting minutes, qualitative overlay documentation, and selected email communications.

During the preparation sessions, Hu was reminded of the Upjohn admonitions — specifically, that Calderwood & Harkness LLP represents BRB the entity, not Hu individually; that communications between Hu and C&H attorneys are protected by the attorney-client privilege belonging to BRB; and that BRB, as the holder of the privilege, may elect to waive that privilege in the future. Hu confirmed his understanding of these principles and acknowledged that he had received the same admonitions at the outset of the engagement in 2023.

Hu was also reminded of the general principles governing SEC investigative testimony, including the requirement to testify truthfully, the right to confer with counsel before answering any question, and the instruction to answer only the question asked without volunteering additional information. Hu demonstrated an excellent understanding of these principles throughout the preparation sessions and during the testimony itself.

III. Summary of Testimony — Key Topics and Q&A

The following is a detailed summary of the testimony organized by substantive topic area. The summary is based on the contemporaneous notes taken by Liam Donahue during the testimony and supplemented by the recollections of Victoria Chambers and Jasmine Thornton. This summary does not purport to be a verbatim transcript and should be supplemented and, where necessary, corrected upon receipt and review of the official transcript.

A. Opening and Background Questions (Led by Andrea Felton)

Senior Staff Attorney Andrea Felton led the opening portion of the testimony, which covered Hu's educational background, professional history, and current role at BRB. This portion lasted approximately forty-five minutes.

Educational and Professional Background. Felton began with standard background questions. Hu confirmed that he holds a Bachelor of Science in Accounting from the University of Virginia (2005) and a Master of Science in Accountancy from the same institution (2006). He is a licensed Certified Public Accountant in the State of New York and the Commonwealth of Virginia, and his licenses are in good standing. He described his fourteen-year career at Whitfield Keane LLP, where he joined as a Staff Auditor in 2005 and departed as a Senior Manager in 2019. He explained that his practice at Whitfield Keane focused on the audit of financial institutions, including commercial banks, savings institutions, and credit unions, and that he developed particular expertise in credit-loss provisioning, fair value accounting, and regulatory capital compliance.

Tenure and Reporting Lines at BRB. Hu confirmed that he has served as Chief Accounting Officer of BRB since June 2019. He reports directly to Patricia Navarrete, Chief Financial Officer, and maintains a dotted-line reporting relationship to Gwendolyn Prescott, Chair of the Audit Committee of BRB's Board of Directors. Hu described the scope of his responsibilities as encompassing the supervision of BRB's general ledger accounting function, the preparation of the Bank's SEC filings (in coordination with outside securities reporting counsel), the design and operation of the CECL model, management of the sub-certification process, and liaison with BRB's independent auditors.

Role in SEC Filings. Felton asked Hu to describe his role in the preparation of BRB's periodic SEC filings — specifically, the FY 2022 Form 10-K (filed February 28, 2023), the Q1 2023 Form 10-Q (filed May 10, 2023), and the Q2 2023 Form 10-Q (filed August 9, 2023). Hu confirmed that he served as the senior officer responsible for the ACL estimation process disclosed in those filings. He explained that his team prepared the ACL calculations, which were then reviewed by the Credit Loss Estimation Committee, approved by Hu, reported to the CFO (Navarrete), and incorporated into the financial statements. Hu confirmed that the credit-loss disclosures appearing in the MD&A and the notes to the financial statements were drafted under his supervision and reviewed by him prior to filing.

Sub-Certification Process. Felton asked Hu about BRB's sub-certification process, a topic she had also explored with Navarrete during the January 18 testimony. Hu explained that, in connection with each periodic filing, he executed a sub-certification attesting to the accuracy and completeness of the financial information within his area of responsibility, including the ACL. He confirmed that his sub-certifications were reviewed and relied upon by the CFO (Navarrete) and the CEO in connection with their certifications under Sections 302 and 906 of the Sarbanes-Oxley Act of 2002. Hu confirmed that he signed sub-certifications for each of the three filings at issue.

Coordination with Outside Counsel. Felton asked whether Hu had coordinated with outside counsel in reviewing the credit-loss disclosures in the filings. Hu confirmed that BRB retained Caldwell Pratt & Stein LLP as its outside securities reporting counsel, and that the lead partner on that engagement was Martin Caldwell. Hu stated that Caldwell Pratt & Stein reviewed drafts of the credit-loss disclosures in the MD&A and the notes to the financial statements prior to each filing. He confirmed that Caldwell Pratt & Stein provided comments on the disclosures and that those comments were incorporated where BRB's management agreed they were appropriate. Hu did not disclose the specific substance of any privileged communications with Caldwell Pratt & Stein.

B. CECL Model Architecture and Inputs (Led by Diane Milosevic)

Following the background segment, Staff Accountant Diane Milosevic assumed the lead questioning role. Milosevic's portion of the testimony constituted the most substantial and technically detailed segment, lasting approximately three hours in total (inclusive of the re-aging discussion addressed in Section III.C below). The nature and specificity of Milosevic's questions reflected a careful and thorough study of BRB's produced documents. As noted above, BRB's document production in this matter has totaled 284,617 documents comprising approximately 1.42 million pages across six rolling productions, the sixth and final of which is scheduled for completion on March 1, 2024. Milosevic appeared to have reviewed a significant volume of this production, including model governance memoranda, Credit Loss Estimation Committee minutes, model validation reports, and underlying model parameter documentation.

Focus Area 1: Model Design. Milosevic asked Hu to describe the overall architecture of BRB's CECL model. Hu provided a detailed explanation of the model's structure. He described the probability-of-default / loss-given-default ("PD/LGD") framework that BRB employs as the core quantitative engine of its CECL model. Under this framework, BRB estimates the probability that a loan will default over its remaining contractual life (adjusted for expected prepayments) and the expected loss severity in the event of default. These two parameters are estimated at the segment level, with loans grouped into homogeneous pools based on shared risk characteristics.

Hu explained that for the BridgeFin portfolio specifically, BRB overlaid the PD/LGD framework with a vintage-cohort segmentation approach. Under this approach, BridgeFin loans are grouped into cohorts based on their origination vintage (i.e., the quarter in which the loan was originated), and PD and LGD parameters are estimated at the cohort level to capture differences in underwriting standards, borrower characteristics, and economic conditions prevailing at the time of origination. Hu stated that this approach was selected because vintage effects are a well-recognized driver of credit performance in consumer lending portfolios, and the BridgeFin portfolio — which experienced rapid growth from its launch in Q3 2020 — was expected to exhibit meaningful vintage variation.

Hu further described the model's use of macroeconomic forecast inputs, including the national unemployment rate, GDP growth rate, and the S&P Harborline Software Case-Shiller Housing Price Index. These macroeconomic variables are incorporated through regression-based relationships estimated from historical data, which translate forecast macroeconomic conditions into projected PD and LGD paths over the reasonable and supportable forecast period (which BRB set at two years for the relevant periods, reverting to long-run historical averages thereafter). Finally, Hu described the qualitative overlay component, which allows management to adjust the model's quantitative output to account for factors not fully captured by the model, such as emerging risks, model limitations, and environmental conditions not reflected in historical data.

Focus Area 2: BridgeFin-Specific Inputs. Milosevic turned to the specific inputs and parameters applicable to the BridgeFin portfolio. She asked Hu to describe the size and composition of the BridgeFin portfolio at the relevant measurement dates. Hu stated that as of December 31, 2022, the BridgeFin portfolio comprised approximately $4.1 billion in outstanding loans. BRB originated approximately $2.87 billion in BridgeFin loans during FY 2022 alone, reflecting the rapid growth trajectory of the platform. The portfolio consisted of two primary loan types: personal loans with an average loan size of approximately $18,400, and micro-business loans with an average loan size of approximately $42,000. Hu noted that BridgeFin was launched in Q3 2020, and accordingly, the oldest loans in the portfolio at December 31, 2022 had a maximum seasoning of approximately two and a half years.

Milosevic explored the limited historical loss data point extensively. She noted that a two-and-a-half-year performance history is short by the standards of credit-loss modeling and asked Hu whether he believed the limited historical dataset created additional estimation uncertainty for the CECL model. Hu acknowledged forthrightly that the limited performance history of the BridgeFin portfolio was a factor that the modeling team considered carefully. He explained, however, that BRB took several steps to mitigate this limitation. First, BRB supplemented the BridgeFin-specific loss data with proxy data drawn from BRB's broader consumer unsecured loan portfolio, which had accumulated over a decade of loss history spanning multiple credit cycles, including the post-2008 period. Hu explained that while the BridgeFin borrower population had certain unique characteristics (e.g., a higher proportion of borrowers in the near-prime and subprime credit tiers), there was sufficient overlap in borrower demographics and loan characteristics to make the proxy data informative. Second, BRB applied qualitative overlays — discussed further in Section III.D below — specifically designed to account for the platform's newer vintage characteristics and the model's inherent uncertainty arising from the limited historical dataset.

Milosevic followed up by asking whether the proxy data adjustment was documented. Hu confirmed that it was, referencing BRB's model methodology document (which he identified by its internal document control number and which had been produced to the Staff in the fourth rolling production). He explained that the model methodology document described the proxy data source, the mapping methodology used to align the proxy data with BridgeFin loan segments, and the adjustments applied to account for known differences in borrower populations.

Focus Area 3: Macroeconomic Forecast Scenarios. Milosevic asked Hu to describe the macroeconomic forecast scenarios used in the CECL model for the FY 2022 year-end measurement date. Hu explained that BRB employed a three-scenario approach: a baseline scenario, an adverse scenario, and a severely adverse scenario. Each scenario was defined by a set of projected macroeconomic variable paths over the two-year reasonable and supportable forecast period. Hu stated that the scenario definitions were informed by third-party economic forecasts and were calibrated by BRB's economics team in consultation with the Credit Loss Estimation Committee.

Milosevic asked for the specific probability weightings BRB assigned to each scenario. Hu provided them, referencing internal model governance documents that had been produced to the Staff. Specifically, Hu stated that for the FY 2022 year-end measurement date, BRB assigned a weighting of 50% to the baseline scenario, 35% to the adverse scenario, and 15% to the severely adverse scenario. Hu explained that these weightings were reviewed and approved by the Credit Loss Estimation Committee and were documented in the committee's meeting minutes for the Q4 2022 meeting. Milosevic asked whether the weightings had changed from prior quarters; Hu stated that the weightings were consistent with those used at the Q3 2022 measurement date but reflected a shift from the Q2 2022 weightings, which had assigned a higher probability (55%) to the baseline scenario. Hu explained that the shift reflected BRB's updated assessment of heightened economic uncertainty in the second half of 2022.

C. Re-Aging / Modification Treatment — Core Disputed Issue (Led by Diane Milosevic)

This section of the testimony consumed the largest single block of time — approximately ninety minutes — and addressed what we understand to be the central issue in the Staff's investigation. Milosevic's questioning was methodical, persistent, and technically precise. Hu remained composed and consistent in his responses throughout.

The Re-Aging Practice. Milosevic asked Hu to describe BRB's treatment of modified and re-aged BridgeFin loans for purposes of the CECL model. Hu explained that BRB's BridgeFin platform offered certain modification options to borrowers experiencing financial hardship, including payment deferrals (typically one to three months), term extensions (typically three to six months), and, in limited cases, interest rate reductions. When a BridgeFin loan was modified under one of these programs, the loan's delinquency clock — i.e., the days-past-due counter that tracks the borrower's payment status — was reset to "current" status in BRB's loan servicing and accounting systems. This practice is commonly referred to as "re-aging."

For CECL modeling purposes, Hu explained, the modified loan was reclassified into a new vintage cohort beginning on the modification date. In other words, the loan was treated as if it had been re-originated on the date of modification, with a new set of contractual cash flows and a fresh performance history. The loan's pre-modification delinquency history was not carried forward into the new vintage assignment for purposes of estimating PD and LGD parameters.

Hu stated that this treatment was BRB's deliberate policy choice, documented in the CECL model methodology and approved by the Credit Loss Estimation Committee. He explained that the rationale was grounded in the view that a modification represents a new credit decision by the Bank and a new set of contractual terms, which fundamentally alters the expected cash flows associated with the loan. In BRB's judgment, assigning the loan to a new vintage was appropriate because the modification changed the loan's risk profile in a manner that was better captured by the characteristics of the new vintage cohort than by the original vintage cohort.

Staff's Concern: Masking Deterioration. Milosevic pressed on this point, asking why BRB did not retain modified loans in their original vintage delinquency buckets for purposes of loss-rate estimation. She stated — framing it as a question — that the Staff's concern was that re-aging effectively masked deteriorating credit quality by removing delinquent loans from the delinquency statistics used as inputs to the CECL model. Milosevic noted that the effect was to make the BridgeFin portfolio appear healthier than it would have appeared had modified loans remained in their original delinquency buckets.

Hu responded that he understood the Staff's concern but disagreed that the treatment masked credit quality. He explained that BRB's treatment was consistent with the guidance in the Office of the Comptroller of the Currency ("OCC") Comptroller's Handbook on troubled debt restructurings and loan modifications, which recognizes that modifications create new contractual terms and may appropriately be reflected as current in a bank's records. He further stated that the modification represented a deliberate credit decision by BRB — the Bank had evaluated the borrower's ability to perform under the modified terms and had concluded that modification was preferable to charge-off or foreclosure — and that treating the modified loan as a new credit was consistent with the economic substance of the transaction.

Hu also emphasized that BRB's modification treatment was disclosed in the Bank's SEC filings. He referenced the MD&A discussion in BRB's FY 2022 Form 10-K, which described BRB's loan modification practices for the BridgeFin portfolio and noted the volume of modifications during the fiscal year. Hu stated that the disclosure was reviewed by Caldwell Pratt & Stein LLP and by Whitfield Keane LLP during the FY 2022 audit.

Quantitative Impact. Milosevic asked whether Hu was aware of the quantitative impact of the re-aging treatment on BRB's reported credit metrics. Specifically, she asked whether Hu was aware that the effect of re-aging reduced BRB's reported 90+ day delinquency rate for the BridgeFin portfolio from what would have been approximately 2.38% — had the re-aged loans been retained in their original delinquency buckets — to the reported figure of 1.72%. Hu acknowledged that he was aware of the mathematical difference. He stated that BRB had performed internal analyses quantifying the effect of the re-aging treatment on the delinquency rate and other credit metrics, and that these analyses were part of the documentation reviewed by the Credit Loss Estimation Committee.

Milosevic then explored the impact on the net charge-off rate, noting that BRB reported a net charge-off rate of 2.14% for the BridgeFin portfolio for FY 2022, whereas the Staff's preliminary view — calculated by including charge-offs on modified loans on an accelerated timeline consistent with the original vintage's expected performance — was that the net charge-off rate should have been approximately 3.05%. Hu explained that charge-offs on modified loans were deferred in timing due to the new contractual terms — i.e., because the modification reset the delinquency clock, a modified loan would not reach charge-off status (typically 120+ days past due) until a later date than it would have absent the modification. He emphasized, however, that the charge-offs were deferred, not eliminated, and that BRB's reported figure of 2.14% reflected actual recognized charge-offs consistent with GAAP as of the measurement date. He stated that BRB's CECL model projected future charge-offs on modified loans based on the modified terms, so the expected lifetime losses were still captured in the ACL — they were simply allocated to different periods than the Staff's alternative methodology would suggest.

Number and Volume of Re-Aged Loans. Milosevic asked Hu to confirm the number and aggregate outstanding balance of re-aged or modified BridgeFin loans. Hu confirmed that the figures were 12,847 loans with an aggregate outstanding balance of approximately $236.4 million as of December 31, 2022. These figures were consistent with data BRB had produced to the Staff in the document production.

Discussion with Whitfield Keane LLP. Milosevic asked whether the re-aging treatment had been specifically discussed with Whitfield Keane LLP during the FY 2022 audit. Hu confirmed that it had been. He stated that audit engagement partner Lorraine Dupont and her team had reviewed BRB's modification and re-aging policy as part of their audit procedures for the ACL. Hu stated that Whitfield Keane had asked questions about the treatment, requested and received supporting documentation (including the model methodology document, Credit Loss Estimation Committee minutes, and internal sensitivity analyses), and had ultimately concluded that BRB's treatment was within the range of acceptable accounting under ASC 326. He stated that Whitfield Keane did not issue any audit findings or require any adjustments to the ACL related to the re-aging treatment. Hu referenced specific audit workpapers by document identifier but was careful not to characterize any privileged communications, and we did not assert privilege on the specific factual question of whether the topic was discussed with the auditors. We note that no privilege was waived.

ACL Impact. Milosevic then turned to the bottom-line ACL impact of the re-aging treatment. She stated that BRB reported an ACL of $143.5 million for the BridgeFin portfolio as of December 31, 2022, representing a coverage ratio of 3.50% of the $4.1 billion in outstanding BridgeFin loans. She stated that the Staff's preliminary view was that the ACL should have been approximately $180.8 million, representing a coverage ratio of 4.41%, resulting in a potential understatement of $37.3 million. She asked Hu whether he agreed with the Staff's calculation.

Hu responded carefully. He stated that he had not independently verified the Staff's specific figure of $180.8 million and could not confirm or deny the precise calculation without reviewing the Staff's assumptions and methodology. However, he acknowledged that if modified loans were retained in their original vintage delinquency buckets — rather than reassigned to new vintage cohorts on the modification date — the model would produce a higher ACL figure. He stated that BRB had itself performed sensitivity analyses at year-end 2022, testing the impact of alternative assumptions including the retention of modified loans in original vintages, and that BRB was aware that alternative assumptions could produce a higher ACL.

Hu then made what we regard as the most important statement on this topic: he testified that even under alternative assumptions, BRB's reported ACL of $143.5 million fell within a defensible range of reasonable estimates under ASC 326. He explained that ASC 326 requires management to estimate expected credit losses over the remaining life of a financial asset, and that this estimation inherently involves significant judgment. He stated that reasonable professionals could disagree on the appropriate treatment of modified loans, the appropriate proxy data adjustments, the macroeconomic scenario weightings, and numerous other assumptions, and that reasonable differences in these inputs could produce a range of ACL estimates, all of which could be considered appropriate under the standard.

Hu further addressed the materiality point. He acknowledged that the $37.3 million difference cited by the Staff exceeded BRB's internal materiality threshold of $30.6 million, which was calculated as 5% of BRB's FY 2022 pre-tax income of $612 million — a difference of $6.7 million above the threshold. However, Hu stated that exceeding the quantitative materiality threshold did not automatically render the reported ACL unreasonable or require a restatement. He explained that materiality is assessed on both a quantitative and qualitative basis, and that BRB considered a number of qualitative factors — including the deliberate and documented nature of the accounting policy, the disclosure of modification practices in the public filings, and the absence of any audit findings from Whitfield Keane — in concluding that the reported ACL was appropriate.

D. Quarterly True-Up Procedures and Qualitative Overlays (Led by Diane Milosevic)

Following the re-aging discussion, Milosevic shifted to the quarterly true-up procedures and qualitative overlay adjustments. This segment lasted approximately thirty minutes.

Quarterly Governance Process. Hu described BRB's quarterly governance process for the CECL model. He explained that, at each quarterly measurement date, the quantitative model was run by BRB's credit analytics team, producing segment-level ACL estimates based on the current loan portfolio, updated macroeconomic forecasts, and refreshed model parameters. The model outputs were then reviewed by BRB's Credit Loss Estimation Committee, which is chaired by Chief Risk Officer Annette Kowalski and includes representatives from credit risk management, the BridgeFin division, the finance group, and Hu's accounting team. The committee met quarterly to review the model outputs, assess the need for qualitative overlay adjustments, and approve the final ACL figures. Following committee approval, the ACL was reviewed and approved by Hu before incorporation into the financial statements.

Hu also described a model validation process. He stated that BRB's internal model risk management team conducts an annual independent validation of the CECL model, assessing the model's conceptual soundness, the appropriateness of its inputs and assumptions, and its backtesting performance against actual credit outcomes. He stated that the most recent validation, conducted in Q3 2022, identified certain areas for potential model enhancement — including an observation that the limited BridgeFin historical data increased estimation uncertainty — but did not identify any issues requiring immediate remediation or that would call into question the model's fitness for use.

Qualitative Overlay Adjustments. Milosevic asked Hu to describe the qualitative overlay adjustments that BRB applied to the CECL model's quantitative output. Hu explained that the qualitative overlay framework was designed to capture factors that the quantitative model might not fully reflect. He listed the specific factors considered by the Credit Loss Estimation Committee, including: (i) emerging delinquency trends not yet reflected in the model's historical loss data; (ii) concentration risk, particularly the rapid growth and high concentration of unsecured consumer and micro-business lending in the BridgeFin portfolio; (iii) economic forecast uncertainty, especially during periods of heightened macroeconomic volatility; (iv) model limitations arising from BridgeFin's limited performance history; and (v) changes in underwriting standards or borrower mix that could affect future credit performance.

Milosevic asked whether qualitative overlays had the effect of increasing or decreasing the modeled ACL for Q1 2023 and Q2 2023. Hu stated that the qualitative overlays added approximately $8 million to $12 million above the model's quantitative output in each of those quarters. He stated that these additions reflected BRB's awareness of potential model limitations — particularly the limited BridgeFin historical loss data — and management's judgment that the quantitative model output alone might not fully capture the portfolio's risk profile. He stated that the magnitude of the overlay was documented in the Credit Loss Estimation Committee minutes for each quarter and was reviewed by Whitfield Keane LLP during their interim review procedures.

Milosevic pressed on whether these overlays were sufficient, given the re-aging issue. She noted that the $8-$12 million overlay was significantly less than the $37.3 million ACL difference the Staff had identified. Hu maintained that the qualitative overlays were reasonable and reflected management's best judgment at the time. He explained that the overlays were intended to address a range of model limitations, not solely the re-aging treatment, and that BRB did not view the re-aging treatment as a "limitation" requiring correction through the overlay process — rather, BRB viewed it as an appropriate accounting policy. He stated that the overlay amounts were calibrated based on a holistic assessment of the factors described above and were within the range of professional judgment.

E. Communications with Whitfield Keane LLP (Led by Andrea Felton)

Andrea Felton resumed the lead questioner role for this segment, which lasted approximately twenty-five minutes.

Cadence of Communications. Felton asked Hu to describe the nature and frequency of his communications with Whitfield Keane LLP regarding the ACL. Hu stated that he had regular quarterly meetings with engagement partner Lorraine Dupont, typically scheduled in the weeks following each quarter-end. These quarterly meetings covered BRB's ACL methodology, the quarter's model outputs, any changes to inputs or assumptions, the qualitative overlay adjustments, and any emerging credit trends. In addition, Hu described year-end deep-dive sessions, typically spanning multiple days, during which the Whitfield Keane audit team conducted a thorough review of the ACL methodology, model validation results, significant management judgments, and the supporting documentation.

Hu confirmed that the BridgeFin ACL was a significant audit area identified by Whitfield Keane in their audit plan, reflecting the portfolio's rapid growth, limited performance history, and the accounting judgments involved. He stated that the audit team devoted substantial attention to the BridgeFin ACL during both the interim review procedures and the year-end audit.

Auditor Awareness of Key Issues. Felton asked whether Whitfield Keane was aware of each of the key issues discussed earlier in the testimony. Hu confirmed the following: (i) Whitfield Keane was aware of the re-aging treatment and the reclassification of modified loans into new vintage cohorts; (ii) Whitfield Keane was aware of the limited historical loss data for the BridgeFin portfolio and BRB's use of proxy data from the broader consumer unsecured loan portfolio; (iii) Whitfield Keane was aware of the qualitative overlay methodology and the magnitude of the overlay adjustments; and (iv) Whitfield Keane was aware of the macroeconomic scenario weightings and the rationale for changes in those weightings over time.

Auditor Conclusions. Felton asked whether Whitfield Keane raised any concerns about the ACL. Hu stated that Whitfield Keane had asked numerous questions and requested additional documentation during the FY 2022 audit. He described this as a normal part of the audit process, particularly for a significant accounting estimate involving substantial management judgment. He stated, however, that Whitfield Keane ultimately concluded that BRB's reported ACL was within an acceptable range of estimates and issued an unqualified (i.e., "clean") audit opinion on BRB's financial statements for FY 2022.

Privilege Instructions. During this segment, Felton asked several questions that touched upon Hu's communications with BRB's outside counsel — both Caldwell Pratt & Stein LLP (securities reporting counsel) and Calderwood & Harkness LLP (defense counsel in this investigation). On each occasion, we instructed Hu not to disclose the specific content of privileged communications with counsel. Felton did not challenge these instructions and moved on to other questions. The privilege instructions were limited and targeted; we did not assert privilege over factual questions regarding the ACL process or communications with the auditors.

F. Internal Controls Over Financial Reporting (Led by Christopher Yuen)

Staff Attorney Christopher Yuen conducted a relatively brief segment — approximately fifteen minutes — focused on BRB's internal controls over financial reporting ("ICFR") as they relate to the BridgeFin ACL.

Design and Effectiveness of Controls. Yuen asked Hu to describe the design and effectiveness of internal controls specific to the BridgeFin ACL. Hu described the following controls:

1. Model Governance Framework. BRB maintains a formal model governance framework, overseen by the Credit Loss Estimation Committee (chaired by Chief Risk Officer Annette Kowalski), that establishes policies and procedures for the development, validation, monitoring, and use of the CECL model. The framework includes requirements for documentation of model assumptions, approval of parameter changes, and periodic backtesting.

2. Independent Model Validation. BRB's internal model risk management team conducts an annual independent validation of the CECL model. The validation is performed by personnel independent of the credit analytics team that builds and operates the model. Validation results are reported to the Credit Loss Estimation Committee and the Audit Committee.

3. Segregation of Duties. Hu described the segregation of duties between BridgeFin loan operations (managed by Naveen Chakravarti's team) and accounting/ACL estimation (managed by Hu's team). He explained that the individuals responsible for originating and servicing BridgeFin loans are different from the individuals responsible for estimating the ACL, and that this segregation is designed to ensure that operational incentives do not influence accounting judgments.

4. Audit Committee Oversight. The Audit Committee, chaired by Gwendolyn Prescott, receives quarterly reports on the ACL methodology, model outputs, qualitative overlays, and any significant changes to assumptions. The Audit Committee also meets periodically with Whitfield Keane LLP without management present.

Hu confirmed that BRB's management assessment of ICFR for FY 2022, as reported in the Form 10-K, identified no material weaknesses. Yuen asked whether BRB identified any significant deficiencies related to the BridgeFin ACL. Hu stated that no significant deficiencies were identified in FY 2022 or in the Q1 or Q2 2023 periods. Yuen did not follow up with any further questions on this topic.

G. Closing and Procedural Matters

Felton resumed for the closing of the testimony. She asked the standard closing questions: whether Hu wished to add anything to or correct any of his testimony, and whether he was aware of any other individuals with information relevant to the investigation who had not yet been identified to the Staff. Hu stated that he did not wish to add or correct anything and that he was not aware of any additional individuals with relevant information who had not already been identified.

The testimony concluded at approximately 4:15 p.m. EST. Calderwood & Harkness reserved the right to review the transcript and submit corrections in accordance with the SEC's standard testimony procedures.

IV. Assessment of SEC Staff's Focus and Understanding

Based on the questioning observed during Hu's testimony, and considered in light of the two prior testimonies, the following assessment of the Staff's investigative focus and current understanding is offered:

Diane Milosevic's Technical Proficiency. Milosevic demonstrated a strong and detailed understanding of ASC 326 / CECL mechanics, vintage-cohort modeling techniques, PD/LGD estimation, macroeconomic forecast integration, and the specific re-aging and modification treatment at issue. Her questions were not generic or template-based; they reflected careful analysis of BRB's produced model documentation, including the ability to identify specific parameter choices, reference particular document production Bates ranges, and calculate the quantitative impact of alternative methodological assumptions. Milosevic's questioning was notably more technically sophisticated than the questioning observed during the Navarrete and Chakravarti testimonies, consistent with the expectation that the Staff would save its most detailed technical questions for the witness most directly responsible for the CECL model. The Staff Accountant appears to be the driving force behind the quantitative theory of the case, and it is clear that she has invested considerable time in understanding BRB's model architecture and the specific choices that gave rise to the alleged ACL understatement.

Andrea Felton's Approach. Felton's questioning was focused on the narrative arc of the investigation — specifically, what did BRB know about deteriorating credit quality in the BridgeFin portfolio, when did it know it, and why didn't it adjust its disclosures or ACL methodology sooner. Felton devoted particular attention to the cadence and substance of communications with Whitfield Keane LLP, suggesting that the Staff may be evaluating whether the auditor interactions create any evidentiary path — either as evidence that BRB was on notice of potential issues, or as a potential basis for challenging the "auditor reliance" element of our defense. However, Felton's questions did not suggest that the Staff has uncovered any smoking-gun communication between BRB and its auditors; her questions appeared exploratory rather than conclusory.

Christopher Yuen's ICFR Questions. Yuen's questions on internal controls appeared preliminary and foundational. They did not suggest that the Staff is building a books-and-records case based on control failures or that the Staff has identified specific control deficiencies. The questions were more likely aimed at establishing a complete record of BRB's governance framework for the ACL, which would be relevant to any enforcement recommendation regardless of whether control failures are alleged. We do not assess Yuen's line of questioning as suggesting an expansion of the Staff's theory.

Overall Staff Posture. The questioning throughout Hu's testimony — and across all three testimonies — remained focused on Sections 13(a) and 13(b)(2)(A) of the Securities Exchange Act of 1934 and Rules 13a-1 and 12b-20 thereunder (periodic reporting accuracy and books and records). The questions were directed at whether BRB's ACL methodology produced materially accurate financial statements and whether the methodology and its effects were adequately disclosed, not at whether BRB's management acted with scienter or an intent to deceive investors. No questions were asked that suggested the Staff is considering scienter-based fraud charges under Section 10(b) of the Exchange Act or Rule 10b-5 thereunder, or that the Staff is investigating any individual officers for personal liability under Section 20(a) or otherwise. This assessment is consistent with the observations documented in the Navarrete and Chakravarti debrief memoranda and, if correct, represents a favorable disposition from BRB's perspective, as the potential remedies and reputational consequences associated with reporting charges are significantly less severe than those associated with fraud charges.

V. Witness Performance and Credibility Assessment

Hu was an excellent witness. He was measured in his pace, precise in his language, and technically fluent in a manner that was authoritative without being condescending or evasive. His fourteen years at Whitfield Keane LLP — including extensive experience with financial institution audits and credit-loss accounting — were clearly evident in his command of the relevant accounting standards, his comfort with audit terminology, and his ability to explain complex CECL model mechanics in a clear and organized fashion. His demeanor was professional and composed throughout, including during the sustained and technically challenging questioning by Milosevic on the re-aging issue.

Hu appropriately deferred on matters outside his direct knowledge or area of responsibility. When asked questions about specific BridgeFin origination decisions, underwriting criteria changes, or operational loan modification protocols, Hu directed the Staff to the testimony of Naveen Chakravarti, who as BridgeFin Division Head was the appropriate witness on those topics. He did so naturally and without appearing evasive or uncooperative.

Consistent with the guidance provided during the preparation sessions, Hu did not volunteer information beyond what was asked. He answered each question directly and then stopped, allowing the Staff to direct the flow of the testimony. He did not speculate, did not offer opinions on matters he had not been asked about, and did not attempt to argue the defense case affirmatively — all of which are consistent with best practices for investigative testimony.

No significant inconsistencies with the prior testimony of either Navarrete or Chakravarti were identified during the testimony or in the post-testimony review of notes. Hu's account of the ACL governance process, the CECL model methodology, the re-aging treatment, the qualitative overlay adjustments, and the interactions with Whitfield Keane LLP was fully consistent with the accounts provided by the other two witnesses, while adding the additional layer of technical detail appropriate to his role as CAO.

On the core re-aging issue — which is the central factual and legal question in the investigation — Hu presented BRB's treatment as a deliberate, documented, and disclosed judgment, not an oversight, error, or manipulation. This framing is critical to the defense narrative. An ACL methodology that was deliberately chosen, documented in model governance materials, disclosed in public filings, and reviewed without objection by an independent auditor is fundamentally different from a methodology that was adopted ad hoc, concealed from auditors, or designed to mask deteriorating credit quality. Hu's testimony effectively established the former characterization.

VI. Comparison with Prior Witness Testimonies

A principal objective in preparing for Hu's testimony was to ensure consistency across the three investigative testimonies. Having now completed all three, we assess the testimonies to be fully consistent and mutually reinforcing. The key points of comparison are as follows:

Patricia Navarrete (January 18, 2024). Navarrete testified to the CFO-level oversight of the ACL estimation process and the certification procedures under the Sarbanes-Oxley Act. She described the flow of information from Hu's team to her office, the review and approval processes for the ACL, and her reliance on Hu's sub-certifications in executing her own certifications. Hu's testimony is fully consistent with Navarrete's account. He confirmed the process Navarrete described and provided the granular technical details underlying her high-level account. Where Navarrete described the "what" (the ACL figures, the governance process, the disclosure review), Hu described the "how" (the model architecture, the parameter choices, the vintage-cohort methodology, the qualitative overlay calibration). There are no contradictions between the two testimonies, and the two accounts are naturally complementary.

Naveen Chakravarti (January 30, 2024). Chakravarti testified to the operational rationale for the BridgeFin loan modification programs, the credit-quality metrics tracked by his division, and the communication of credit risk information from the BridgeFin division to the accounting team. Hu's testimony is likewise consistent with and complementary to Chakravarti's account. Where Chakravarti explained the business rationale for modifications (i.e., that modifications were offered to borrowers who demonstrated a willingness to repay and whose financial circumstances were expected to improve, and that modification was preferable to charge-off from both a customer-relationship and a loss-mitigation perspective), Hu explained how those operational decisions were translated into accounting treatment in the CECL model. Hu's description of the segregation of duties between Chakravarti's operations team and Hu's accounting team was consistent with Chakravarti's own description of the organizational structure. No contradictions were identified.

Combined Narrative. Taken together, the three testimonies present a coherent and mutually reinforcing narrative: BridgeFin's modification practices were operationally justified and consistent with sound credit risk management (Chakravarti); the accounting treatment of those modifications was properly implemented under a reasonable CECL methodology, deliberate, documented, and disclosed (Hu); and the ACL figures were appropriately overseen, reviewed, and certified by senior management (Navarrete). This narrative is the foundation of the defense position and should be reflected in the Wells Submission.

VII. Implications for Defense Strategy

Hu's testimony reinforces and strengthens the core defense position that BRB's reported ACL of $143.5 million for the BridgeFin portfolio as of December 31, 2022, reflected a reasonable exercise of management judgment under ASC 326, even if the Staff believes that a higher figure — specifically, $180.8 million, or $37.3 million more — would have been more appropriate. The defense does not need to prove that BRB's ACL was the single "correct" answer; it needs to demonstrate that it was within the range of reasonable estimates, a standard that is inherent in the CECL framework and well-recognized in SEC enforcement precedent.

Re-Aging as the Central Battleground. The re-aging treatment remains the central factual and legal issue in the investigation. Milosevic's detailed questioning on the quantitative impact of re-aging — the $37.3 million ACL difference, the 2.14% versus 3.05% net charge-off rate, and the 1.72% versus 2.38% 90+ day delinquency rate — confirms that the Staff is focused on this specific methodology choice as the primary, and perhaps sole, driver of the alleged understatement. Hu's testimony that this treatment was (a) deliberately chosen by BRB's management based on a documented policy rationale, (b) reflected in formal model governance documentation approved by the Credit Loss Estimation Committee, (c) disclosed in the MD&A of BRB's FY 2022 Form 10-K and subsequent filings, and (d) reviewed without objection by Whitfield Keane LLP during the FY 2022 audit provides strong factual support for the defense position that the treatment was reasonable and in good faith.

Materiality. The materiality question will be important. Hu's testimony that the $37.3 million difference exceeded BRB's internal quantitative materiality threshold of $30.6 million by only $6.7 million, and that BRB considered qualitative factors in concluding that the reported ACL was nonetheless appropriate, provides a factual basis for arguing that any potential misstatement was, at most, at the margin of materiality. The Wells Submission should develop this argument further, emphasizing the well-established principle that a quantitative threshold is only a starting point and that qualitative factors must also be considered (see SAB 99; Litwin v. Stonefield Private Capital, 634 F.3d 706 (2d Cir. 2011)).

Expert Report. In light of Milosevic's technically sophisticated questioning, we recommend that the expert report being prepared by Dr. Franklin Osei of Brindoval Analytic Partners directly address the range-of-estimates argument with detailed sensitivity analyses. Specifically, the report should demonstrate that BRB's reported ACL falls within a reasonable range of estimates under ASC 326 even under modified assumptions for the treatment of re-aged loans. The report should present multiple scenarios — ranging from full retention of re-aged loans in original vintage cohorts to partial retention with varying cure-rate assumptions — and show the resulting ACL figures. The report should also include a peer comparison analysis, comparing BRB's ACL coverage ratio for the BridgeFin portfolio against coverage ratios reported by peer digital lending institutions for similar portfolios, to demonstrate that BRB's ratio was not an outlier.

Anticipated Charges. Based on Hu's testimony and the overall tenor of the investigation across all three testimonies, we continue to assess that any enforcement action would likely be limited to the reporting provisions of the Securities Exchange Act of 1934 — specifically, Section 13(a) (periodic reporting requirements), Section 13(b)(2)(A) (books and records requirements), and Rules 13a-1 and 12b-20 thereunder — rather than the anti-fraud provisions under Section 10(b) and Rule 10b-5. The absence of any scienter-focused questioning across all three testimonies strongly supports this assessment. This is a materially favorable posture for BRB, as the potential sanctions, litigation risk, and reputational consequences associated with reporting charges are significantly less severe than those associated with fraud charges. Nevertheless, we should prepare the Wells Submission to address both categories in the event that the Staff's recommendation is broader than anticipated.

Wells Call. We should anticipate a Wells call in the coming weeks, likely following the completion of BRB's final document production on March 1, 2024 and the Staff's review of the Hu transcript. The Wells Submission should be substantially drafted and ready for finalization promptly upon receipt of the Wells call.

VIII. Open Items and Next Steps

The following items require attention in the near term:

1. Transcript Review. Request and review the transcript of Hu's testimony upon availability from the court reporter. Compare the transcript against Liam Donahue's contemporaneous notes for any discrepancies or areas where corrections may be appropriate. Submit any necessary corrections in accordance with the SEC's standard transcript correction procedures. Responsible: Liam Donahue. Target: Upon receipt of transcript.

2. Final Document Production. Finalize the sixth and final rolling document production to the SEC, which is scheduled for completion on March 1, 2024. The final production will encompass approximately 38,912 documents. Coordinate with BRB's e-discovery vendor (Axion Data Solutions) and BRB's internal IT and legal teams to ensure timely completion. Responsible: Liam Donahue (coordination); Jasmine Thornton (review and sign-off). Target: March 1, 2024.

3. Interim Status Memorandum to Client. Prepare an interim status memorandum for Audit Committee Chair Gwendolyn Prescott and General Counsel Derek Sung, incorporating findings and assessments from all three investigative testimonies (Navarrete, Chakravarti, Hu). The memorandum should provide a consolidated assessment of the Staff's investigative focus, the consistency of the testimony record, the anticipated enforcement posture, and recommended next steps. Responsible: Jasmine Thornton (draft); Victoria Chambers (review and approval). Target: February 22, 2024.

4. Brindoval Analytic Partners Expert Report. Continue coordinating with Brindoval Analytic Partners (Dr. Franklin Osei) on the forensic accounting analysis and expert report. In light of Milosevic's questioning, the expert engagement should place particular emphasis on: (a) the vintage-cohort sensitivity analysis demonstrating BRB's ACL within a reasonable range of estimates under alternative re-aging assumptions; (b) a peer comparison analysis of ACL coverage ratios among peer digital lending institutions; and (c) a critique of the Staff's preliminary ACL calculation of $180.8 million, identifying any methodological assumptions in the Staff's analysis that are themselves debatable. Responsible: Jasmine Thornton (day-to-day coordination); Victoria Chambers (strategic direction). Target: Draft report by March 15, 2024.

5. Wells Submission Preparation. Continue preparation of the Wells Submission in parallel. The submission should be substantially drafted and ready for finalization promptly upon receipt of the anticipated Wells call. The submission should incorporate the key defense themes reinforced by the three testimonies, the Brindoval expert analysis, and any additional arguments identified through ongoing research and analysis. Responsible: Jasmine Thornton and Liam Donahue (drafting); Victoria Cha

PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT PREPARED IN ANTICIPATION OF LITIGATION NOT TO BE DISCLOSED WITHOUT PRIOR AUTHORIZATION
CALDERWOOD & HARKNESS LLP 425 Lexington Avenue, 30th Floor New York, New York 10017
February 15, 2024
MEMORANDUM
TO: Victoria Chambers, Partner Howard Quinn, Partner
FROM: Jasmine Thornton, Lead Associate
CC: Liam Donahue, Working Associate
RE: Post-Testimony Debrief — Investigative Testimony of Raymond Hu, Chief Accounting Officer, Belmont Ridge Bank NA (February 14, 2024)
MATTER: SEC Investigation File No. HO-14287 C&H Engagement No. SC-MAT-2023-4172
PRIVILEGE LEGEND: This memorandum is a privileged and confidential attorney work product of Calderwood & Harkness LLP. It is protected in its entirety by the attorney-client privilege and the attorney work product doctrine. It was prepared in anticipation of litigation and reflects the mental impressions, conclusions, opinions, and legal theories of counsel. This memorandum is not to be disclosed, distributed, forwarded, copied, or otherwise shared outside the Calderwood & Harkness LLP engagement team and authorized Belmont Ridge Bank NA personnel without the prior written approval of Victoria Chambers.
I. Executive Summary
Raymond Hu, Chief Accounting Officer of Belmont Ridge Bank National Association ("BRB" or the "Bank"), testified under oath on February 14, 2024, before staff of the United States Securities and Exchange Commission ("SEC") Division of Enforcement, pursuant to the Formal Order of Investigation No. EN-2023-0947 (the "Formal Order"). The testimony was conducted at the SEC Division of Enforcement offices located at 100 F Street NE, Washington, D.C. 20549, and lasted approximately five and a half hours, inclusive of a mid-morning break and a lunch recess.
Hu's testimony was the third and final scheduled investigative testimony in this matter. The prior two witnesses testified as follows: (i) Patricia Navarrete, Chief Financial Officer of BRB, testified on January 18, 2024; and (ii) Naveen Chakravarti, Head of BRB's BridgeFin Division, testified on January 30, 2024. With the completion of Hu's testimony, the Staff has now examined the three principal BRB officers most directly involved in the matters under investigation — namely, BRB's accounting for the allowance for credit losses ("ACL") on the BridgeFin loan portfolio under ASC 326 (the Current Expected Credit Losses, or "CECL," standard) and related financial reporting disclosures.
Hu was represented at the testimony by Victoria Chambers (Partner) and Jasmine Thornton (Lead Associate) of Calderwood & Harkness LLP. Liam Donahue (Working Associate) attended in a supporting capacity for note-taking and document management. SEC Staff present at the testimony included Senior Staff Attorney Andrea Felton, who served as the lead questioner for approximately the first two hours; Staff Attorney Christopher Yuen, who asked a discrete set of questions concerning internal controls over financial reporting; and Staff Accountant Diane Milosevic, who served as the lead questioner for approximately three hours and focused on the technical mechanics of BRB's CECL model, with particular emphasis on the treatment of re-aged and modified BridgeFin loans.
Overall Assessment: Hu performed well throughout the testimony and presented as credible, methodical, and precise. His extensive background in public accounting — including fourteen years at Whitfield Keane LLP, BRB's independent auditor — was evident in his facility with accounting standards, audit terminology, and the granular mechanics of CECL model construction. He maintained composure and consistency even under sustained and technically detailed questioning from Milosevic, whose line of inquiry was notably more granular and technically sophisticated than we had anticipated. Milosevic's questioning suggests that the Staff has developed a deep and well-informed understanding of BRB's CECL model mechanics, likely aided by an outside accounting consultant or by Milosevic's own considerable expertise. Notwithstanding the rigor of the Staff's questioning, Hu's testimony effectively conveyed that BRB's CECL methodology was deliberate, well-documented, consistent with applicable accounting standards, and within the range of reasonable estimates under ASC 326. No significant inconsistencies with the prior testimonies of Navarrete or Chakravarti were identified. The testimony reinforces the core defense position for the anticipated Wells Submission.
II. Witness Background and Preparation
A. Professional Background
Raymond Hu has served as Chief Accounting Officer of BRB since June 2019. Prior to joining BRB, Hu spent fourteen years at Whitfield Keane LLP (PCAOB Registration No. 2847), BRB's independent auditor, where he held positions of increasing responsibility within the firm's audit practice. During his tenure at Whitfield Keane, Hu rose to the level of Senior Manager and was involved in the audits of several large regional and community banking institutions, although he was not a member of the BRB engagement team during his time at the firm. His decision to transition to the client side was contemporaneous with BRB's strategic expansion into digital lending through the BridgeFin platform, and he was hired in part for his deep expertise in credit-loss accounting and financial institution regulatory reporting.
At BRB, Hu reports directly to CFO Patricia Navarrete. He also maintains a dotted-line reporting relationship to Gwendolyn Prescott, Chair of BRB's Audit Committee. In his capacity as CAO, Hu was the primary BRB officer responsible for the design, implementation, and ongoing operation of the Bank's CECL model following BRB's adoption of ASC 326, which became effective for BRB on January 1, 2020 (its first fiscal year beginning after December 15, 2019, as required for SEC filers). He also supervised the preparation of the credit-loss disclosures appearing in BRB's periodic SEC filings, including the Management's Discussion and Analysis ("MD&A") and notes to the financial statements addressing the ACL.
B. Preparation Sessions
Witness preparation sessions were conducted on February 9, 11, and 13, 2024, at Calderwood & Harkness's New York offices. The sessions totaled approximately fourteen hours across the three days. The preparation team consisted of Victoria Chambers, Jasmine Thornton, and Liam Donahue.
The substance of the preparation sessions covered the following principal areas:
1. CECL model architecture and documentation — A comprehensive review of BRB's CECL model, including the probability-of-default / loss-given-default ("PD/LGD") framework, vintage-cohort segmentation methodology, and the specific parameters and assumptions applied to the BridgeFin portfolio.
2. Vintage cohort treatment and re-aging/modification practices — Detailed discussion of BRB's accounting treatment for modified and re-aged BridgeFin loans, including the reassignment of modified loans to new vintage cohorts, the rationale for this treatment, and the documentary support in BRB's model governance materials.
3. Quarterly true-up procedures and qualitative overlay adjustments — Review of BRB's process for quarterly recalibration of the CECL model, including the role of the Credit Loss Estimation Committee, the qualitative overlay factors applied, and the magnitude of overlay adjustments in the relevant periods.
4. Communications with Whitfield Keane LLP — Discussion of the cadence and substance of Hu's interactions with Whitfield Keane LLP engagement partner Lorraine Dupont and her audit team, particularly with respect to the FY 2022 year-end audit and the interim reviews for Q1 and Q2 2023.
5. Anticipated SEC focus areas — Identification of likely Staff focus areas based on the questioning observed during the Navarrete testimony (January 18, 2024) and the Chakravarti testimony (January 30, 2024), including the re-aging issue, the limited historical loss data for BridgeFin, the macroeconomic scenario weightings, and the potential materiality of the ACL difference.
6. Document familiarity — Review of key documents from BRB's production that we anticipated the Staff would use as exhibits during the testimony, including model governance memoranda, Credit Loss Estimation Committee meeting minutes, qualitative overlay documentation, and selected email communications.
During the preparation sessions, Hu was reminded of the Upjohn admonitions — specifically, that Calderwood & Harkness LLP represents BRB the entity, not Hu individually; that communications between Hu and C&H attorneys are protected by the attorney-client privilege belonging to BRB; and that BRB, as the holder of the privilege, may elect to waive that privilege in the future. Hu confirmed his understanding of these principles and acknowledged that he had received the same admonitions at the outset of the engagement in 2023.
Hu was also reminded of the general principles governing SEC investigative testimony, including the requirement to testify truthfully, the right to confer with counsel before answering any question, and the instruction to answer only the question asked without volunteering additional information. Hu demonstrated an excellent understanding of these principles throughout the preparation sessions and during the testimony itself.
III. Summary of Testimony — Key Topics and Q&A
The following is a detailed summary of the testimony organized by substantive topic area. The summary is based on the contemporaneous notes taken by Liam Donahue during the testimony and supplemented by the recollections of Victoria Chambers and Jasmine Thornton. This summary does not purport to be a verbatim transcript and should be supplemented and, where necessary, corrected upon receipt and review of the official transcript.
A. Opening and Background Questions (Led by Andrea Felton)
Senior Staff Attorney Andrea Felton led the opening portion of the testimony, which covered Hu's educational background, professional history, and current role at BRB. This portion lasted approximately forty-five minutes.
Educational and Professional Background. Felton began with standard background questions. Hu confirmed that he holds a Bachelor of Science in Accounting from the University of Virginia (2005) and a Master of Science in Accountancy from the same institution (2006). He is a licensed Certified Public Accountant in the State of New York and the Commonwealth of Virginia, and his licenses are in good standing. He described his fourteen-year career at Whitfield Keane LLP, where he joined as a Staff Auditor in 2005 and departed as a Senior Manager in 2019. He explained that his practice at Whitfield Keane focused on the audit of financial institutions, including commercial banks, savings institutions, and credit unions, and that he developed particular expertise in credit-loss provisioning, fair value accounting, and regulatory capital compliance.
Tenure and Reporting Lines at BRB. Hu confirmed that he has served as Chief Accounting Officer of BRB since June 2019. He reports directly to Patricia Navarrete, Chief Financial Officer, and maintains a dotted-line reporting relationship to Gwendolyn Prescott, Chair of the Audit Committee of BRB's Board of Directors. Hu described the scope of his responsibilities as encompassing the supervision of BRB's general ledger accounting function, the preparation of the Bank's SEC filings (in coordination with outside securities reporting counsel), the design and operation of the CECL model, management of the sub-certification process, and liaison with BRB's independent auditors.
Role in SEC Filings. Felton asked Hu to describe his role in the preparation of BRB's periodic SEC filings — specifically, the FY 2022 Form 10-K (filed February 28, 2023), the Q1 2023 Form 10-Q (filed May 10, 2023), and the Q2 2023 Form 10-Q (filed August 9, 2023). Hu confirmed that he served as the senior officer responsible for the ACL estimation process disclosed in those filings. He explained that his team prepared the ACL calculations, which were then reviewed by the Credit Loss Estimation Committee, approved by Hu, reported to the CFO (Navarrete), and incorporated into the financial statements. Hu confirmed that the credit-loss disclosures appearing in the MD&A and the notes to the financial statements were drafted under his supervision and reviewed by him prior to filing.
Sub-Certification Process. Felton asked Hu about BRB's sub-certification process, a topic she had also explored with Navarrete during the January 18 testimony. Hu explained that, in connection with each periodic filing, he executed a sub-certification attesting to the accuracy and completeness of the financial information within his area of responsibility, including the ACL. He confirmed that his sub-certifications were reviewed and relied upon by the CFO (Navarrete) and the CEO in connection with their certifications under Sections 302 and 906 of the Sarbanes-Oxley Act of 2002. Hu confirmed that he signed sub-certifications for each of the three filings at issue.
Coordination with Outside Counsel. Felton asked whether Hu had coordinated with outside counsel in reviewing the credit-loss disclosures in the filings. Hu confirmed that BRB retained Caldwell Pratt & Stein LLP as its outside securities reporting counsel, and that the lead partner on that engagement was Martin Caldwell. Hu stated that Caldwell Pratt & Stein reviewed drafts of the credit-loss disclosures in the MD&A and the notes to the financial statements prior to each filing. He confirmed that Caldwell Pratt & Stein provided comments on the disclosures and that those comments were incorporated where BRB's management agreed they were appropriate. Hu did not disclose the specific substance of any privileged communications with Caldwell Pratt & Stein.
B. CECL Model Architecture and Inputs (Led by Diane Milosevic)
Following the background segment, Staff Accountant Diane Milosevic assumed the lead questioning role. Milosevic's portion of the testimony constituted the most substantial and technically detailed segment, lasting approximately three hours in total (inclusive of the re-aging discussion addressed in Section III.C below). The nature and specificity of Milosevic's questions reflected a careful and thorough study of BRB's produced documents. As noted above, BRB's document production in this matter has totaled 284,617 documents comprising approximately 1.42 million pages across six rolling productions, the sixth and final of which is scheduled for completion on March 1, 2024. Milosevic appeared to have reviewed a significant volume of this production, including model governance memoranda, Credit Loss Estimation Committee minutes, model validation reports, and underlying model parameter documentation.
Focus Area 1: Model Design. Milosevic asked Hu to describe the overall architecture of BRB's CECL model. Hu provided a detailed explanation of the model's structure. He described the probability-of-default / loss-given-default ("PD/LGD") framework that BRB employs as the core quantitative engine of its CECL model. Under this framework, BRB estimates the probability that a loan will default over its remaining contractual life (adjusted for expected prepayments) and the expected loss severity in the event of default. These two parameters are estimated at the segment level, with loans grouped into homogeneous pools based on shared risk characteristics.
Hu explained that for the BridgeFin portfolio specifically, BRB overlaid the PD/LGD framework with a vintage-cohort segmentation approach. Under this approach, BridgeFin loans are grouped into cohorts based on their origination vintage (i.e., the quarter in which the loan was originated), and PD and LGD parameters are estimated at the cohort level to capture differences in underwriting standards, borrower characteristics, and economic conditions prevailing at the time of origination. Hu stated that this approach was selected because vintage effects are a well-recognized driver of credit performance in consumer lending portfolios, and the BridgeFin portfolio — which experienced rapid growth from its launch in Q3 2020 — was expected to exhibit meaningful vintage variation.
Hu further described the model's use of macroeconomic forecast inputs, including the national unemployment rate, GDP growth rate, and the S&P Harborline Software Case-Shiller Housing Price Index. These macroeconomic variables are incorporated through regression-based relationships estimated from historical data, which translate forecast macroeconomic conditions into projected PD and LGD paths over the reasonable and supportable forecast period (which BRB set at two years for the relevant periods, reverting to long-run historical averages thereafter). Finally, Hu described the qualitative overlay component, which allows management to adjust the model's quantitative output to account for factors not fully captured by the model, such as emerging risks, model limitations, and environmental conditions not reflected in historical data.
Focus Area 2: BridgeFin-Specific Inputs. Milosevic turned to the specific inputs and parameters applicable to the BridgeFin portfolio. She asked Hu to describe the size and composition of the BridgeFin portfolio at the relevant measurement dates. Hu stated that as of December 31, 2022, the BridgeFin portfolio comprised approximately $4.1 billion in outstanding loans. BRB originated approximately $2.87 billion in BridgeFin loans during FY 2022 alone, reflecting the rapid growth trajectory of the platform. The portfolio consisted of two primary loan types: personal loans with an average loan size of approximately $18,400, and micro-business loans with an average loan size of approximately $42,000. Hu noted that BridgeFin was launched in Q3 2020, and accordingly, the oldest loans in the portfolio at December 31, 2022 had a maximum seasoning of approximately two and a half years.
Milosevic explored the limited historical loss data point extensively. She noted that a two-and-a-half-year performance history is short by the standards of credit-loss modeling and asked Hu whether he believed the limited historical dataset created additional estimation uncertainty for the CECL model. Hu acknowledged forthrightly that the limited performance history of the BridgeFin portfolio was a factor that the modeling team considered carefully. He explained, however, that BRB took several steps to mitigate this limitation. First, BRB supplemented the BridgeFin-specific loss data with proxy data drawn from BRB's broader consumer unsecured loan portfolio, which had accumulated over a decade of loss history spanning multiple credit cycles, including the post-2008 period. Hu explained that while the BridgeFin borrower population had certain unique characteristics (e.g., a higher proportion of borrowers in the near-prime and subprime credit tiers), there was sufficient overlap in borrower demographics and loan characteristics to make the proxy data informative. Second, BRB applied qualitative overlays — discussed further in Section III.D below — specifically designed to account for the platform's newer vintage characteristics and the model's inherent uncertainty arising from the limited historical dataset.
Milosevic followed up by asking whether the proxy data adjustment was documented. Hu confirmed that it was, referencing BRB's model methodology document (which he identified by its internal document control number and which had been produced to the Staff in the fourth rolling production). He explained that the model methodology document described the proxy data source, the mapping methodology used to align the proxy data with BridgeFin loan segments, and the adjustments applied to account for known differences in borrower populations.
Focus Area 3: Macroeconomic Forecast Scenarios. Milosevic asked Hu to describe the macroeconomic forecast scenarios used in the CECL model for the FY 2022 year-end measurement date. Hu explained that BRB employed a three-scenario approach: a baseline scenario, an adverse scenario, and a severely adverse scenario. Each scenario was defined by a set of projected macroeconomic variable paths over the two-year reasonable and supportable forecast period. Hu stated that the scenario definitions were informed by third-party economic forecasts and were calibrated by BRB's economics team in consultation with the Credit Loss Estimation Committee.
Milosevic asked for the specific probability weightings BRB assigned to each scenario. Hu provided them, referencing internal model governance documents that had been produced to the Staff. Specifically, Hu stated that for the FY 2022 year-end measurement date, BRB assigned a weighting of 50% to the baseline scenario, 35% to the adverse scenario, and 15% to the severely adverse scenario. Hu explained that these weightings were reviewed and approved by the Credit Loss Estimation Committee and were documented in the committee's meeting minutes for the Q4 2022 meeting. Milosevic asked whether the weightings had changed from prior quarters; Hu stated that the weightings were consistent with those used at the Q3 2022 measurement date but reflected a shift from the Q2 2022 weightings, which had assigned a higher probability (55%) to the baseline scenario. Hu explained that the shift reflected BRB's updated assessment of heightened economic uncertainty in the second half of 2022.
C. Re-Aging / Modification Treatment — Core Disputed Issue (Led by Diane Milosevic)
This section of the testimony consumed the largest single block of time — approximately ninety minutes — and addressed what we understand to be the central issue in the Staff's investigation. Milosevic's questioning was methodical, persistent, and technically precise. Hu remained composed and consistent in his responses throughout.
The Re-Aging Practice. Milosevic asked Hu to describe BRB's treatment of modified and re-aged BridgeFin loans for purposes of the CECL model. Hu explained that BRB's BridgeFin platform offered certain modification options to borrowers experiencing financial hardship, including payment deferrals (typically one to three months), term extensions (typically three to six months), and, in limited cases, interest rate reductions. When a BridgeFin loan was modified under one of these programs, the loan's delinquency clock — i.e., the days-past-due counter that tracks the borrower's payment status — was reset to "current" status in BRB's loan servicing and accounting systems. This practice is commonly referred to as "re-aging."
For CECL modeling purposes, Hu explained, the modified loan was reclassified into a new vintage cohort beginning on the modification date. In other words, the loan was treated as if it had been re-originated on the date of modification, with a new set of contractual cash flows and a fresh performance history. The loan's pre-modification delinquency history was not carried forward into the new vintage assignment for purposes of estimating PD and LGD parameters.
Hu stated that this treatment was BRB's deliberate policy choice, documented in the CECL model methodology and approved by the Credit Loss Estimation Committee. He explained that the rationale was grounded in the view that a modification represents a new credit decision by the Bank and a new set of contractual terms, which fundamentally alters the expected cash flows associated with the loan. In BRB's judgment, assigning the loan to a new vintage was appropriate because the modification changed the loan's risk profile in a manner that was better captured by the characteristics of the new vintage cohort than by the original vintage cohort.
Staff's Concern: Masking Deterioration. Milosevic pressed on this point, asking why BRB did not retain modified loans in their original vintage delinquency buckets for purposes of loss-rate estimation. She stated — framing it as a question — that the Staff's concern was that re-aging effectively masked deteriorating credit quality by removing delinquent loans from the delinquency statistics used as inputs to the CECL model. Milosevic noted that the effect was to make the BridgeFin portfolio appear healthier than it would have appeared had modified loans remained in their original delinquency buckets.
Hu responded that he understood the Staff's concern but disagreed that the treatment masked credit quality. He explained that BRB's treatment was consistent with the guidance in the Office of the Comptroller of the Currency ("OCC") Comptroller's Handbook on troubled debt restructurings and loan modifications, which recognizes that modifications create new contractual terms and may appropriately be reflected as current in a bank's records. He further stated that the modification represented a deliberate credit decision by BRB — the Bank had evaluated the borrower's ability to perform under the modified terms and had concluded that modification was preferable to charge-off or foreclosure — and that treating the modified loan as a new credit was consistent with the economic substance of the transaction.
Hu also emphasized that BRB's modification treatment was disclosed in the Bank's SEC filings. He referenced the MD&A discussion in BRB's FY 2022 Form 10-K, which described BRB's loan modification practices for the BridgeFin portfolio and noted the volume of modifications during the fiscal year. Hu stated that the disclosure was reviewed by Caldwell Pratt & Stein LLP and by Whitfield Keane LLP during the FY 2022 audit.
Quantitative Impact. Milosevic asked whether Hu was aware of the quantitative impact of the re-aging treatment on BRB's reported credit metrics. Specifically, she asked whether Hu was aware that the effect of re-aging reduced BRB's reported 90+ day delinquency rate for the BridgeFin portfolio from what would have been approximately 2.38% — had the re-aged loans been retained in their original delinquency buckets — to the reported figure of 1.72%. Hu acknowledged that he was aware of the mathematical difference. He stated that BRB had performed internal analyses quantifying the effect of the re-aging treatment on the delinquency rate and other credit metrics, and that these analyses were part of the documentation reviewed by the Credit Loss Estimation Committee.
Milosevic then explored the impact on the net charge-off rate, noting that BRB reported a net charge-off rate of 2.14% for the BridgeFin portfolio for FY 2022, whereas the Staff's preliminary view — calculated by including charge-offs on modified loans on an accelerated timeline consistent with the original vintage's expected performance — was that the net charge-off rate should have been approximately 3.05%. Hu explained that charge-offs on modified loans were deferred in timing due to the new contractual terms — i.e., because the modification reset the delinquency clock, a modified loan would not reach charge-off status (typically 120+ days past due) until a later date than it would have absent the modification. He emphasized, however, that the charge-offs were deferred, not eliminated, and that BRB's reported figure of 2.14% reflected actual recognized charge-offs consistent with GAAP as of the measurement date. He stated that BRB's CECL model projected future charge-offs on modified loans based on the modified terms, so the expected lifetime losses were still captured in the ACL — they were simply allocated to different periods than the Staff's alternative methodology would suggest.
Number and Volume of Re-Aged Loans. Milosevic asked Hu to confirm the number and aggregate outstanding balance of re-aged or modified BridgeFin loans. Hu confirmed that the figures were 12,847 loans with an aggregate outstanding balance of approximately $236.4 million as of December 31, 2022. These figures were consistent with data BRB had produced to the Staff in the document production.
Discussion with Whitfield Keane LLP. Milosevic asked whether the re-aging treatment had been specifically discussed with Whitfield Keane LLP during the FY 2022 audit. Hu confirmed that it had been. He stated that audit engagement partner Lorraine Dupont and her team had reviewed BRB's modification and re-aging policy as part of their audit procedures for the ACL. Hu stated that Whitfield Keane had asked questions about the treatment, requested and received supporting documentation (including the model methodology document, Credit Loss Estimation Committee minutes, and internal sensitivity analyses), and had ultimately concluded that BRB's treatment was within the range of acceptable accounting under ASC 326. He stated that Whitfield Keane did not issue any audit findings or require any adjustments to the ACL related to the re-aging treatment. Hu referenced specific audit workpapers by document identifier but was careful not to characterize any privileged communications, and we did not assert privilege on the specific factual question of whether the topic was discussed with the auditors. We note that no privilege was waived.
ACL Impact. Milosevic then turned to the bottom-line ACL impact of the re-aging treatment. She stated that BRB reported an ACL of $143.5 million for the BridgeFin portfolio as of December 31, 2022, representing a coverage ratio of 3.50% of the $4.1 billion in outstanding BridgeFin loans. She stated that the Staff's preliminary view was that the ACL should have been approximately $180.8 million, representing a coverage ratio of 4.41%, resulting in a potential understatement of $37.3 million. She asked Hu whether he agreed with the Staff's calculation.
Hu responded carefully. He stated that he had not independently verified the Staff's specific figure of $180.8 million and could not confirm or deny the precise calculation without reviewing the Staff's assumptions and methodology. However, he acknowledged that if modified loans were retained in their original vintage delinquency buckets — rather than reassigned to new vintage cohorts on the modification date — the model would produce a higher ACL figure. He stated that BRB had itself performed sensitivity analyses at year-end 2022, testing the impact of alternative assumptions including the retention of modified loans in original vintages, and that BRB was aware that alternative assumptions could produce a higher ACL.
Hu then made what we regard as the most important statement on this topic: he testified that even under alternative assumptions, BRB's reported ACL of $143.5 million fell within a defensible range of reasonable estimates under ASC 326. He explained that ASC 326 requires management to estimate expected credit losses over the remaining life of a financial asset, and that this estimation inherently involves significant judgment. He stated that reasonable professionals could disagree on the appropriate treatment of modified loans, the appropriate proxy data adjustments, the macroeconomic scenario weightings, and numerous other assumptions, and that reasonable differences in these inputs could produce a range of ACL estimates, all of which could be considered appropriate under the standard.
Hu further addressed the materiality point. He acknowledged that the $37.3 million difference cited by the Staff exceeded BRB's internal materiality threshold of $30.6 million, which was calculated as 5% of BRB's FY 2022 pre-tax income of $612 million — a difference of $6.7 million above the threshold. However, Hu stated that exceeding the quantitative materiality threshold did not automatically render the reported ACL unreasonable or require a restatement. He explained that materiality is assessed on both a quantitative and qualitative basis, and that BRB considered a number of qualitative factors — including the deliberate and documented nature of the accounting policy, the disclosure of modification practices in the public filings, and the absence of any audit findings from Whitfield Keane — in concluding that the reported ACL was appropriate.
D. Quarterly True-Up Procedures and Qualitative Overlays (Led by Diane Milosevic)
Following the re-aging discussion, Milosevic shifted to the quarterly true-up procedures and qualitative overlay adjustments. This segment lasted approximately thirty minutes.
Quarterly Governance Process. Hu described BRB's quarterly governance process for the CECL model. He explained that, at each quarterly measurement date, the quantitative model was run by BRB's credit analytics team, producing segment-level ACL estimates based on the current loan portfolio, updated macroeconomic forecasts, and refreshed model parameters. The model outputs were then reviewed by BRB's Credit Loss Estimation Committee, which is chaired by Chief Risk Officer Annette Kowalski and includes representatives from credit risk management, the BridgeFin division, the finance group, and Hu's accounting team. The committee met quarterly to review the model outputs, assess the need for qualitative overlay adjustments, and approve the final ACL figures. Following committee approval, the ACL was reviewed and approved by Hu before incorporation into the financial statements.
Hu also described a model validation process. He stated that BRB's internal model risk management team conducts an annual independent validation of the CECL model, assessing the model's conceptual soundness, the appropriateness of its inputs and assumptions, and its backtesting performance against actual credit outcomes. He stated that the most recent validation, conducted in Q3 2022, identified certain areas for potential model enhancement — including an observation that the limited BridgeFin historical data increased estimation uncertainty — but did not identify any issues requiring immediate remediation or that would call into question the model's fitness for use.
Qualitative Overlay Adjustments. Milosevic asked Hu to describe the qualitative overlay adjustments that BRB applied to the CECL model's quantitative output. Hu explained that the qualitative overlay framework was designed to capture factors that the quantitative model might not fully reflect. He listed the specific factors considered by the Credit Loss Estimation Committee, including: (i) emerging delinquency trends not yet reflected in the model's historical loss data; (ii) concentration risk, particularly the rapid growth and high concentration of unsecured consumer and micro-business lending in the BridgeFin portfolio; (iii) economic forecast uncertainty, especially during periods of heightened macroeconomic volatility; (iv) model limitations arising from BridgeFin's limited performance history; and (v) changes in underwriting standards or borrower mix that could affect future credit performance.
Milosevic asked whether qualitative overlays had the effect of increasing or decreasing the modeled ACL for Q1 2023 and Q2 2023. Hu stated that the qualitative overlays added approximately $8 million to $12 million above the model's quantitative output in each of those quarters. He stated that these additions reflected BRB's awareness of potential model limitations — particularly the limited BridgeFin historical loss data — and management's judgment that the quantitative model output alone might not fully capture the portfolio's risk profile. He stated that the magnitude of the overlay was documented in the Credit Loss Estimation Committee minutes for each quarter and was reviewed by Whitfield Keane LLP during their interim review procedures.
Milosevic pressed on whether these overlays were sufficient, given the re-aging issue. She noted that the $8-$12 million overlay was significantly less than the $37.3 million ACL difference the Staff had identified. Hu maintained that the qualitative overlays were reasonable and reflected management's best judgment at the time. He explained that the overlays were intended to address a range of model limitations, not solely the re-aging treatment, and that BRB did not view the re-aging treatment as a "limitation" requiring correction through the overlay process — rather, BRB viewed it as an appropriate accounting policy. He stated that the overlay amounts were calibrated based on a holistic assessment of the factors described above and were within the range of professional judgment.
E. Communications with Whitfield Keane LLP (Led by Andrea Felton)
Andrea Felton resumed the lead questioner role for this segment, which lasted approximately twenty-five minutes.
Cadence of Communications. Felton asked Hu to describe the nature and frequency of his communications with Whitfield Keane LLP regarding the ACL. Hu stated that he had regular quarterly meetings with engagement partner Lorraine Dupont, typically scheduled in the weeks following each quarter-end. These quarterly meetings covered BRB's ACL methodology, the quarter's model outputs, any changes to inputs or assumptions, the qualitative overlay adjustments, and any emerging credit trends. In addition, Hu described year-end deep-dive sessions, typically spanning multiple days, during which the Whitfield Keane audit team conducted a thorough review of the ACL methodology, model validation results, significant management judgments, and the supporting documentation.
Hu confirmed that the BridgeFin ACL was a significant audit area identified by Whitfield Keane in their audit plan, reflecting the portfolio's rapid growth, limited performance history, and the accounting judgments involved. He stated that the audit team devoted substantial attention to the BridgeFin ACL during both the interim review procedures and the year-end audit.
Auditor Awareness of Key Issues. Felton asked whether Whitfield Keane was aware of each of the key issues discussed earlier in the testimony. Hu confirmed the following: (i) Whitfield Keane was aware of the re-aging treatment and the reclassification of modified loans into new vintage cohorts; (ii) Whitfield Keane was aware of the limited historical loss data for the BridgeFin portfolio and BRB's use of proxy data from the broader consumer unsecured loan portfolio; (iii) Whitfield Keane was aware of the qualitative overlay methodology and the magnitude of the overlay adjustments; and (iv) Whitfield Keane was aware of the macroeconomic scenario weightings and the rationale for changes in those weightings over time.
Auditor Conclusions. Felton asked whether Whitfield Keane raised any concerns about the ACL. Hu stated that Whitfield Keane had asked numerous questions and requested additional documentation during the FY 2022 audit. He described this as a normal part of the audit process, particularly for a significant accounting estimate involving substantial management judgment. He stated, however, that Whitfield Keane ultimately concluded that BRB's reported ACL was within an acceptable range of estimates and issued an unqualified (i.e., "clean") audit opinion on BRB's financial statements for FY 2022.
Privilege Instructions. During this segment, Felton asked several questions that touched upon Hu's communications with BRB's outside counsel — both Caldwell Pratt & Stein LLP (securities reporting counsel) and Calderwood & Harkness LLP (defense counsel in this investigation). On each occasion, we instructed Hu not to disclose the specific content of privileged communications with counsel. Felton did not challenge these instructions and moved on to other questions. The privilege instructions were limited and targeted; we did not assert privilege over factual questions regarding the ACL process or communications with the auditors.
F. Internal Controls Over Financial Reporting (Led by Christopher Yuen)
Staff Attorney Christopher Yuen conducted a relatively brief segment — approximately fifteen minutes — focused on BRB's internal controls over financial reporting ("ICFR") as they relate to the BridgeFin ACL.
Design and Effectiveness of Controls. Yuen asked Hu to describe the design and effectiveness of internal controls specific to the BridgeFin ACL. Hu described the following controls:
1. Model Governance Framework. BRB maintains a formal model governance framework, overseen by the Credit Loss Estimation Committee (chaired by Chief Risk Officer Annette Kowalski), that establishes policies and procedures for the development, validation, monitoring, and use of the CECL model. The framework includes requirements for documentation of model assumptions, approval of parameter changes, and periodic backtesting.
2. Independent Model Validation. BRB's internal model risk management team conducts an annual independent validation of the CECL model. The validation is performed by personnel independent of the credit analytics team that builds and operates the model. Validation results are reported to the Credit Loss Estimation Committee and the Audit Committee.
3. Segregation of Duties. Hu described the segregation of duties between BridgeFin loan operations (managed by Naveen Chakravarti's team) and accounting/ACL estimation (managed by Hu's team). He explained that the individuals responsible for originating and servicing BridgeFin loans are different from the individuals responsible for estimating the ACL, and that this segregation is designed to ensure that operational incentives do not influence accounting judgments.
4. Audit Committee Oversight. The Audit Committee, chaired by Gwendolyn Prescott, receives quarterly reports on the ACL methodology, model outputs, qualitative overlays, and any significant changes to assumptions. The Audit Committee also meets periodically with Whitfield Keane LLP without management present.
Hu confirmed that BRB's management assessment of ICFR for FY 2022, as reported in the Form 10-K, identified no material weaknesses. Yuen asked whether BRB identified any significant deficiencies related to the BridgeFin ACL. Hu stated that no significant deficiencies were identified in FY 2022 or in the Q1 or Q2 2023 periods. Yuen did not follow up with any further questions on this topic.
G. Closing and Procedural Matters
Felton resumed for the closing of the testimony. She asked the standard closing questions: whether Hu wished to add anything to or correct any of his testimony, and whether he was aware of any other individuals with information relevant to the investigation who had not yet been identified to the Staff. Hu stated that he did not wish to add or correct anything and that he was not aware of any additional individuals with relevant information who had not already been identified.
The testimony concluded at approximately 4:15 p.m. EST. Calderwood & Harkness reserved the right to review the transcript and submit corrections in accordance with the SEC's standard testimony procedures.
IV. Assessment of SEC Staff's Focus and Understanding
Based on the questioning observed during Hu's testimony, and considered in light of the two prior testimonies, the following assessment of the Staff's investigative focus and current understanding is offered:
Diane Milosevic's Technical Proficiency. Milosevic demonstrated a strong and detailed understanding of ASC 326 / CECL mechanics, vintage-cohort modeling techniques, PD/LGD estimation, macroeconomic forecast integration, and the specific re-aging and modification treatment at issue. Her questions were not generic or template-based; they reflected careful analysis of BRB's produced model documentation, including the ability to identify specific parameter choices, reference particular document production Bates ranges, and calculate the quantitative impact of alternative methodological assumptions. Milosevic's questioning was notably more technically sophisticated than the questioning observed during the Navarrete and Chakravarti testimonies, consistent with the expectation that the Staff would save its most detailed technical questions for the witness most directly responsible for the CECL model. The Staff Accountant appears to be the driving force behind the quantitative theory of the case, and it is clear that she has invested considerable time in understanding BRB's model architecture and the specific choices that gave rise to the alleged ACL understatement.
Andrea Felton's Approach. Felton's questioning was focused on the narrative arc of the investigation — specifically, what did BRB know about deteriorating credit quality in the BridgeFin portfolio, when did it know it, and why didn't it adjust its disclosures or ACL methodology sooner. Felton devoted particular attention to the cadence and substance of communications with Whitfield Keane LLP, suggesting that the Staff may be evaluating whether the auditor interactions create any evidentiary path — either as evidence that BRB was on notice of potential issues, or as a potential basis for challenging the "auditor reliance" element of our defense. However, Felton's questions did not suggest that the Staff has uncovered any smoking-gun communication between BRB and its auditors; her questions appeared exploratory rather than conclusory.
Christopher Yuen's ICFR Questions. Yuen's questions on internal controls appeared preliminary and foundational. They did not suggest that the Staff is building a books-and-records case based on control failures or that the Staff has identified specific control deficiencies. The questions were more likely aimed at establishing a complete record of BRB's governance framework for the ACL, which would be relevant to any enforcement recommendation regardless of whether control failures are alleged. We do not assess Yuen's line of questioning as suggesting an expansion of the Staff's theory.
Overall Staff Posture. The questioning throughout Hu's testimony — and across all three testimonies — remained focused on Sections 13(a) and 13(b)(2)(A) of the Securities Exchange Act of 1934 and Rules 13a-1 and 12b-20 thereunder (periodic reporting accuracy and books and records). The questions were directed at whether BRB's ACL methodology produced materially accurate financial statements and whether the methodology and its effects were adequately disclosed, not at whether BRB's management acted with scienter or an intent to deceive investors. No questions were asked that suggested the Staff is considering scienter-based fraud charges under Section 10(b) of the Exchange Act or Rule 10b-5 thereunder, or that the Staff is investigating any individual officers for personal liability under Section 20(a) or otherwise. This assessment is consistent with the observations documented in the Navarrete and Chakravarti debrief memoranda and, if correct, represents a favorable disposition from BRB's perspective, as the potential remedies and reputational consequences associated with reporting charges are significantly less severe than those associated with fraud charges.
V. Witness Performance and Credibility Assessment
Hu was an excellent witness. He was measured in his pace, precise in his language, and technically fluent in a manner that was authoritative without being condescending or evasive. His fourteen years at Whitfield Keane LLP — including extensive experience with financial institution audits and credit-loss accounting — were clearly evident in his command of the relevant accounting standards, his comfort with audit terminology, and his ability to explain complex CECL model mechanics in a clear and organized fashion. His demeanor was professional and composed throughout, including during the sustained and technically challenging questioning by Milosevic on the re-aging issue.
Hu appropriately deferred on matters outside his direct knowledge or area of responsibility. When asked questions about specific BridgeFin origination decisions, underwriting criteria changes, or operational loan modification protocols, Hu directed the Staff to the testimony of Naveen Chakravarti, who as BridgeFin Division Head was the appropriate witness on those topics. He did so naturally and without appearing evasive or uncooperative.
Consistent with the guidance provided during the preparation sessions, Hu did not volunteer information beyond what was asked. He answered each question directly and then stopped, allowing the Staff to direct the flow of the testimony. He did not speculate, did not offer opinions on matters he had not been asked about, and did not attempt to argue the defense case affirmatively — all of which are consistent with best practices for investigative testimony.
No significant inconsistencies with the prior testimony of either Navarrete or Chakravarti were identified during the testimony or in the post-testimony review of notes. Hu's account of the ACL governance process, the CECL model methodology, the re-aging treatment, the qualitative overlay adjustments, and the interactions with Whitfield Keane LLP was fully consistent with the accounts provided by the other two witnesses, while adding the additional layer of technical detail appropriate to his role as CAO.
On the core re-aging issue — which is the central factual and legal question in the investigation — Hu presented BRB's treatment as a deliberate, documented, and disclosed judgment, not an oversight, error, or manipulation. This framing is critical to the defense narrative. An ACL methodology that was deliberately chosen, documented in model governance materials, disclosed in public filings, and reviewed without objection by an independent auditor is fundamentally different from a methodology that was adopted ad hoc, concealed from auditors, or designed to mask deteriorating credit quality. Hu's testimony effectively established the former characterization.
VI. Comparison with Prior Witness Testimonies
A principal objective in preparing for Hu's testimony was to ensure consistency across the three investigative testimonies. Having now completed all three, we assess the testimonies to be fully consistent and mutually reinforcing. The key points of comparison are as follows:
Patricia Navarrete (January 18, 2024). Navarrete testified to the CFO-level oversight of the ACL estimation process and the certification procedures under the Sarbanes-Oxley Act. She described the flow of information from Hu's team to her office, the review and approval processes for the ACL, and her reliance on Hu's sub-certifications in executing her own certifications. Hu's testimony is fully consistent with Navarrete's account. He confirmed the process Navarrete described and provided the granular technical details underlying her high-level account. Where Navarrete described the "what" (the ACL figures, the governance process, the disclosure review), Hu described the "how" (the model architecture, the parameter choices, the vintage-cohort methodology, the qualitative overlay calibration). There are no contradictions between the two testimonies, and the two accounts are naturally complementary.
Naveen Chakravarti (January 30, 2024). Chakravarti testified to the operational rationale for the BridgeFin loan modification programs, the credit-quality metrics tracked by his division, and the communication of credit risk information from the BridgeFin division to the accounting team. Hu's testimony is likewise consistent with and complementary to Chakravarti's account. Where Chakravarti explained the business rationale for modifications (i.e., that modifications were offered to borrowers who demonstrated a willingness to repay and whose financial circumstances were expected to improve, and that modification was preferable to charge-off from both a customer-relationship and a loss-mitigation perspective), Hu explained how those operational decisions were translated into accounting treatment in the CECL model. Hu's description of the segregation of duties between Chakravarti's operations team and Hu's accounting team was consistent with Chakravarti's own description of the organizational structure. No contradictions were identified.
Combined Narrative. Taken together, the three testimonies present a coherent and mutually reinforcing narrative: BridgeFin's modification practices were operationally justified and consistent with sound credit risk management (Chakravarti); the accounting treatment of those modifications was properly implemented under a reasonable CECL methodology, deliberate, documented, and disclosed (Hu); and the ACL figures were appropriately overseen, reviewed, and certified by senior management (Navarrete). This narrative is the foundation of the defense position and should be reflected in the Wells Submission.
VII. Implications for Defense Strategy
Hu's testimony reinforces and strengthens the core defense position that BRB's reported ACL of $143.5 million for the BridgeFin portfolio as of December 31, 2022, reflected a reasonable exercise of management judgment under ASC 326, even if the Staff believes that a higher figure — specifically, $180.8 million, or $37.3 million more — would have been more appropriate. The defense does not need to prove that BRB's ACL was the single "correct" answer; it needs to demonstrate that it was within the range of reasonable estimates, a standard that is inherent in the CECL framework and well-recognized in SEC enforcement precedent.
Re-Aging as the Central Battleground. The re-aging treatment remains the central factual and legal issue in the investigation. Milosevic's detailed questioning on the quantitative impact of re-aging — the $37.3 million ACL difference, the 2.14% versus 3.05% net charge-off rate, and the 1.72% versus 2.38% 90+ day delinquency rate — confirms that the Staff is focused on this specific methodology choice as the primary, and perhaps sole, driver of the alleged understatement. Hu's testimony that this treatment was (a) deliberately chosen by BRB's management based on a documented policy rationale, (b) reflected in formal model governance documentation approved by the Credit Loss Estimation Committee, (c) disclosed in the MD&A of BRB's FY 2022 Form 10-K and subsequent filings, and (d) reviewed without objection by Whitfield Keane LLP during the FY 2022 audit provides strong factual support for the defense position that the treatment was reasonable and in good faith.
Materiality. The materiality question will be important. Hu's testimony that the $37.3 million difference exceeded BRB's internal quantitative materiality threshold of $30.6 million by only $6.7 million, and that BRB considered qualitative factors in concluding that the reported ACL was nonetheless appropriate, provides a factual basis for arguing that any potential misstatement was, at most, at the margin of materiality. The Wells Submission should develop this argument further, emphasizing the well-established principle that a quantitative threshold is only a starting point and that qualitative factors must also be considered (see SAB 99; Litwin v. Stonefield Private Capital, 634 F.3d 706 (2d Cir. 2011)).
Expert Report. In light of Milosevic's technically sophisticated questioning, we recommend that the expert report being prepared by Dr. Franklin Osei of Brindoval Analytic Partners directly address the range-of-estimates argument with detailed sensitivity analyses. Specifically, the report should demonstrate that BRB's reported ACL falls within a reasonable range of estimates under ASC 326 even under modified assumptions for the treatment of re-aged loans. The report should present multiple scenarios — ranging from full retention of re-aged loans in original vintage cohorts to partial retention with varying cure-rate assumptions — and show the resulting ACL figures. The report should also include a peer comparison analysis, comparing BRB's ACL coverage ratio for the BridgeFin portfolio against coverage ratios reported by peer digital lending institutions for similar portfolios, to demonstrate that BRB's ratio was not an outlier.
Anticipated Charges. Based on Hu's testimony and the overall tenor of the investigation across all three testimonies, we continue to assess that any enforcement action would likely be limited to the reporting provisions of the Securities Exchange Act of 1934 — specifically, Section 13(a) (periodic reporting requirements), Section 13(b)(2)(A) (books and records requirements), and Rules 13a-1 and 12b-20 thereunder — rather than the anti-fraud provisions under Section 10(b) and Rule 10b-5. The absence of any scienter-focused questioning across all three testimonies strongly supports this assessment. This is a materially favorable posture for BRB, as the potential sanctions, litigation risk, and reputational consequences associated with reporting charges are significantly less severe than those associated with fraud charges. Nevertheless, we should prepare the Wells Submission to address both categories in the event that the Staff's recommendation is broader than anticipated.
Wells Call. We should anticipate a Wells call in the coming weeks, likely following the completion of BRB's final document production on March 1, 2024 and the Staff's review of the Hu transcript. The Wells Submission should be substantially drafted and ready for finalization promptly upon receipt of the Wells call.
VIII. Open Items and Next Steps
The following items require attention in the near term:
1. Transcript Review. Request and review the transcript of Hu's testimony upon availability from the court reporter. Compare the transcript against Liam Donahue's contemporaneous notes for any discrepancies or areas where corrections may be appropriate. Submit any necessary corrections in accordance with the SEC's standard transcript correction procedures. Responsible: Liam Donahue. Target: Upon receipt of transcript.
2. Final Document Production. Finalize the sixth and final rolling document production to the SEC, which is scheduled for completion on March 1, 2024. The final production will encompass approximately 38,912 documents. Coordinate with BRB's e-discovery vendor (Axion Data Solutions) and BRB's internal IT and legal teams to ensure timely completion. Responsible: Liam Donahue (coordination); Jasmine Thornton (review and sign-off). Target: March 1, 2024.
3. Interim Status Memorandum to Client. Prepare an interim status memorandum for Audit Committee Chair Gwendolyn Prescott and General Counsel Derek Sung, incorporating findings and assessments from all three investigative testimonies (Navarrete, Chakravarti, Hu). The memorandum should provide a consolidated assessment of the Staff's investigative focus, the consistency of the testimony record, the anticipated enforcement posture, and recommended next steps. Responsible: Jasmine Thornton (draft); Victoria Chambers (review and approval). Target: February 22, 2024.
4. Brindoval Analytic Partners Expert Report. Continue coordinating with Brindoval Analytic Partners (Dr. Franklin Osei) on the forensic accounting analysis and expert report. In light of Milosevic's questioning, the expert engagement should place particular emphasis on: (a) the vintage-cohort sensitivity analysis demonstrating BRB's ACL within a reasonable range of estimates under alternative re-aging assumptions; (b) a peer comparison analysis of ACL coverage ratios among peer digital lending institutions; and (c) a critique of the Staff's preliminary ACL calculation of $180.8 million, identifying any methodological assumptions in the Staff's analysis that are themselves debatable. Responsible: Jasmine Thornton (day-to-day coordination); Victoria Chambers (strategic direction). Target: Draft report by March 15, 2024.
5. Wells Submission Preparation. Continue preparation of the Wells Submission in parallel. The submission should be substantially drafted and ready for finalization promptly upon receipt of the anticipated Wells call. The submission should incorporate the key defense themes reinforced by the three testimonies, the Brindoval expert analysis, and any additional arguments identified through ongoing research and analysis. Responsible: Jasmine Thornton and Liam Donahue (drafting); Victoria Cha

MEMORANDUM

Record

Parties

Dates

Amounts