BOV to Loan Balance Methodology

Scope

This page owns the reusable definitions, evidence hierarchy, dataset construction, and bias controls for BOV-to-loan-balance work. Empirical vintage, asset-class, and resolved-loan findings live in BOVs as Percent of Loan Balance Findings 2026.

BOV as a percentage of loan balance is a distressed-credit screening ratio, not an appraisal, recovery estimate, or market-wide loss rate. Match BOV date, loan balance date, collateral scope, and asset class; separate original, current, senior, and total debt; and compare vintages only under a consistent source definition.

Metric Definitions

BOV to Loan Balance

BOV as % of loan balance = broker opinion of value / current principal balance

This ratio answers: "If a broker's current value opinion were realized as sale proceeds before costs, what percentage of the loan principal would it cover?"

Appraisal to Loan Balance

Updated appraisal as % of loan balance = most recent appraised value / current principal balance

This is the closest widely observable substitute. In public CMBS reporting, updated appraisals are more commonly disclosed than BOVs.

Implied Current LTV

Implied current LTV = current principal balance / BOV or updated appraised value

This is the inverse of the coverage ratio. A value/balance ratio of 80% equals an implied current LTV of 125%.

Appraisal Decline to Value/Balance Proxy

When a source reports only a value decline from origination, the rough translation is:

Current value / current balance ~= (1 - value decline from origination) / original LTV

This is only an approximation. It works best for interest-only loans where current balance is close to original balance. It overstates coverage if total exposure includes advances, default interest, fees, unpaid taxes, insurance, legal costs, or liquidation fees. It understates coverage if the loan amortized materially.

ARA to Implied Value/Balance Proxy

ARA is not a BOV. It is a CMBS waterfall and advancing mechanism. It remains useful as a distress proxy because the calculation is tied to the most recent collateral value. In simplified form:

ARA ~= exposure - (90% * most recent appraised value + eligible reserves)

If current principal balance approximates exposure and reserves are ignored:

Implied value / balance ~= (1 - ARA as % of balance) / 0.90

That rough math turns a 70% ARA into an implied value/balance of about 33%. It should not be read as a final loss forecast.

Evidence Hierarchy

Normalized Dataset Build

The public record is enough to define the model and build a targeted resolved-loan panel, but not enough to fill a complete normalized panel by asset class and quarter. A serious paid-data or servicer-file build should use the following structure.

The data-build path is no longer "find the right public report." It is "assemble a public-remittance, servicer-file, or paid-data panel." Fitch's Torchlight servicer report is the clearest public workflow clue: within 90 days of a transfer to special servicing or REO conversion, the servicer obtains an updated appraisal and at least one BOV. Those value marks exist operationally, but public articles usually expose them only when a loan becomes newsworthy, a rating agency cites it, or a remittance/loss report includes enough fields to infer the ratio.

Public SEC and Remittance Path

There is a practical no-account path for a targeted public sample:

1. Use SEC company submissions to identify the issuer CIK and monthly 10-D / ABS-EE filings for a trust.
2. Use ABS-EE Schedule AL XML for repeatable loan/property fields: current scheduled or actual balance, property type, origination/securitization date, original valuation, most recent valuation amount/date/source when populated, special-servicing dates, advances, and realized loss fields.
3. Use the 10-D Exhibit 99.1 distribution report when available for the richer CREFC-style tables. In BMARK 2018-B2's March 2025 filing, the exhibit's "Historical Liquidated Loan Detail" table included beginning scheduled balance, most recent appraised value or BPO, proceeds, fees / advances / expenses, net proceeds received, net proceeds available for distribution, realized loss, and loss percentage.
4. Join the public 10-D exhibit rows back to ABS-EE and prospectus records by prospectus ID / loan number, then normalize field names across deal administrators.
5. Classify rows by valuation source where disclosed. CREFC IRP distinguishes most recent valuation amount, valuation date, and valuation source; CREFC modification templates also contemplate appraisal and BOV collateral-valuation fields.

The first local implementation is scripts/extract_sec_cmbs_remittance_bov_panel.py, with per-trust outputs under 2026-05-28-sec-edgar-cmbs-remittance-bov-panel* and compact CSV copies under sec_edgar_cmbs_remittance_bov_panel. The combined panel is built by scripts/aggregate_sec_cmbs_remittance_bov_panel.py.

Current combined outputs:

• combined_liquidated_loan_rows
• combined_liquidated_loan_latest_by_loan
• combined_specially_serviced_loan_rows
• combined_specially_serviced_loan_detail_rows
• combined_summary

The public path can expose the numerator, denominator, proceeds, expense, and realized-loss fields for specific trusts and months. Resolved-loan output needs separate principal-balance coverage and total-exposure/net-recovery denominator families. Active special-servicing ratios also need valuation-scope controls because pari passu, component, and portfolio loans can share a whole-collateral appraisal.

The current panel results, vintage mix, and asset-class labeling inventory are maintained in BOVs as Percent of Loan Balance Findings 2026.

Account and Paid-Data Tiers

No account is needed for SEC EDGAR, public KBRA press releases, public Morningstar DBRS press releases, public Trepp articles, or public CRED iQ articles. Accounts or subscriptions become necessary when the research goal shifts from a targeted public sample to a complete repeatable panel.

Required Loan-Level Fields

Core Calculations

bov_balance_pct = BOV_value / current_principal_balance
appraisal_balance_pct = appraised_value / current_principal_balance
value_exposure_pct = appraised_value / total_exposure
current_ltv = current_principal_balance / appraised_value
exposure_ltv = total_exposure / appraised_value
ara_balance_pct = ARA_amount / current_principal_balance
implied_value_balance_from_ara = (1 - ara_balance_pct) / 0.90
origination_value_decline = 1 - appraised_value / original_appraised_value

Grouping

The most useful cuts are:

• origination vintage by year: 2012, 2013, 2014, etc.
• vintage cohort: pre-2015, 2015-2016, 2017-2018, 2019, 2020, 2021-2022, 2023+
• asset class and subtype
• appraisal or BOV quarter
• maturity year
• loan status at valuation date
• geography and market tier
• fixed-rate versus floating-rate
• conduit versus SASB versus CRE CLO

Bias Controls

Do not average raw ratios without controlling for selection. BOVs and updated appraisals are more likely to be ordered when something is wrong. A BOV time series built only from special-serviced loans is a stress-cohort time series, not a market-wide valuation index.

Minimum controls:

• separate special-serviced loans from performing loans
• separate maturity defaults from payment defaults
• winsorize extreme ratios or report medians and quartiles
• flag appraisal vintage versus origination vintage
• show both principal-balance coverage and total-exposure coverage
• split retail malls from necessity retail and office trophy from commodity office
• keep agency multifamily performance cohorts separate from private-label CMBS, SASB, CRE CLO, and bridge-loan valuation cohorts
• separate appraisal/BOV marks from final realized loss

Related Pages

• BOVs as Percent of Loan Balance Findings 2026
• CMBS and Commercial Mortgage Securitization
• CRE Valuation Methodologies
• Distressed Asset Underwriting
• Analyses Hub

Sources

• Source Stack: BOV and CMBS Value-to-Loan-Balance Research 2026
• Source Collection: Trepp Research & Insights Public Crawl 2026