Non-QM Bank Statement Mortgage: An Overview
What is a Non-QM Mortgage?
A Non-Qualified Mortgage (Non-QM) is a mortgage product that falls outside the regulatory definitions of a “Qualified Mortgage” under CFPB rules, meaning it allows more flexible underwriting and documentation standards.
These loans are often used by borrowers who have valid income but cannot present it via standard formats (W-2s, pay stubs) — e.g. self-employed, freelancers, gig workers, 1099 earners, or real estate investors.
Among Non-QM classes, Bank Statement Loans are a popular variant: borrowers present 12–24 months of bank statements (personal &/or business), and lenders analyze cash flow or deposits to derive qualifying income.
Key features and tradeoffs:
| Feature | Advantage | Tradeoff / Risk |
|---|---|---|
| Flexible documentation | Enables borrowers who lack W-2s or traditional proof of income | More manual underwriting, higher scrutiny |
| Broader borrower pool | Self-employed, commission-only, part-time | Higher interest rates, larger down payments |
| Custom underwriting logic | Holistic view of borrower | Requires more data, expert judgment |
| Risk to lender | Potential for higher default risk | Need for stronger validation and controls |
Designing a Bank Statement Non-QM Product
When a lender (or mortgage bank) designs a Non-QM Bank Statement product, consider the following:
- Documentation Requirements
- 12 or 24 months of bank statements (business, personal, or both)
- Identification, credit, asset statements
- (Optional) supplemental documents (signed profit & loss statement, invoices)
- Explanation of large deposits or unusual transactions
- Income Calculation Methodology
- Gross deposits method: Sum all deposits and deduct non-qualifying items (transfers, refunds, internal transfers)
- Adjusted net income method: After deductions for business expenses
- Recurring deposits detection: Identify stable sources (salary, contracts) — separate from one-time lumps
- Trend or declining income flags: Review last 3–6 months for downward trend, which may reduce qualifying income
- Debt-to-Income (DTI) or Residual DTI
Because borrowers may have varied income, flexible DTI thresholds (e.g. up to 45–50%) may be used, especially for strong compensating factors (high reserves, low LTV, good credit). - Reserves & Down Payment
Non-QM products typically require more reserves (e.g. 6–12 months payments) and higher down payments (10–20%+ depending on credit). - Pricing & Spreads
Interest rates are typically 0.5% to several percentage points above conventional QM rates to compensate for underwriting and credit risk. - Risk Controls & Exceptions
- Flagging large deposits, frequent overdrafts
- Requiring explanations or supporting documentation
- Use of overlays (e.g. credit score minimum, maximum exposure limits)
- Reserve buffers or escrow holdbacks
- Secondary Market / Investor Guidelines
Define the investor acceptance criteria (if the loans are to be sold). Many secondary markets now accept Non-QM with strict documentation and eligibility requirements.
Implementation in Encompass
To support a Non-QM Bank Statement product within Encompass, you must configure the system to handle the product’s unique underwriting features.
Key steps:
- Product / Pricing Setup
- Define a new Loan Product (e.g. “NonQM BankStmt 12M”, “NonQM BankStmt 24M”)
- Specify rate tiers, spreads, pricing algorithms, eligibility rules (credit, LTV, DTI, reserves)
- Custom Fields & Data Capture
- Add custom fields to store fields like “Total Deposits 12M”, “Non-qualifying Deposit Deductions”, “Average Monthly Income (BankStmt)”
- Add flags or indicators (e.g. “Large Deposit Flag”, “Trend Decline Flag”)
- Use validation logic or business rules in Encompass to enforce thresholds (e.g. DTI limit, reserves)
- Document Upload & Workflows
- Define document groups: Bank Statements, Source of Funds, Explanations
- Create workflows/tasks to request missing documents, send reminders, assign review
- Integration with Document AI / OCR / Income Calculators
- Integrate Encompass with third-party AI/OCR tools (e.g. Ocrolus) to automatically extract, categorize, validate bank statement line items.
- Pass parsed results into the custom fields above
- Use business rule engines in Encompass to assess acceptability vs thresholds
- Underwriting Automation & Decision Engines
- Incorporate rule engine logic: e.g. if (AvgIncome ≥ threshold AND DTI ≤ limit AND no major flags) pass, else escalate
- Optionally integrate AI models for scoring & risk ranking before underwriter review
- Use Encompass Conditions & automated alerts for unusual items
- Audit & Exception Management
- Maintain audit trails for any manual overrides
- Flag exceptions for underwriter review
- Store rationale for large deposit exceptions, trend declines, compensating factors
- Reporting & Monitoring
- Build dashboards in Encompass to monitor Non-QM pipeline, flagged conditions, conversion rates
- Monitor performance / default metrics for feedback into pricing and guidelines
- Training & Documentation
- Train underwriters and processors on how to read AI-extracted statement reports
- Document product guidelines, exceptions, and decision rules
By building Encompass this way, you can streamline the origination, underwriting, and monitoring of Non-QM Bank Statement loans with consistency and scale.
How AI Enhances Bank Statement Non-QM Products
AI (machine learning, NLP, computer vision) can play a powerful role in making Non-QM Bank Statement lending more efficient, accurate, and scalable. Below are key use cases and implementation ideas.
1. Automated Bank Statement Analysis & Income Parsing
- OCR & Document Parsing
Use AI/OCR engines to read PDF or image bank statements and extract transaction-level data (date, description, amounts). - Classification / Categorization
Classify deposit vs withdrawal, distinguish internal transfers, reversals, refunds, non-income deposits. - Recurring Income Detection & Trend Analysis
Identify stable or recurring deposits (e.g. monthly salary, client payments) vs one-time lumps. Flag declining trends. - Anomaly Detection & Fraud Flagging
Spot suspicious large deposits, frequent overdrafts, closed accounts, or inconsistent patterns. - Income Calculation Engines
Automatically compute average qualifying income using investor rules (e.g. ignoring non-qualifying line items) and populate LOS fields.
Tools like Ocrolus’s Bank Statement Income Calculator are built for this.
2. Automated Underwriting & Decision Engines
- Rule + ML hybrid engines
Combine lender’s rule engine (thresholds, overlays) with ML predictive models that score risk or likelihood of default. - Compensating Factor Scoring
Let AI models weigh compensating factors (reserves, credit history, LTV buffer) to approve borderline cases. - Scenario Simulation & “What-If” analysis
Run multiple product scenarios, altering terms or rate, to find optimal fits for borrowers.
3. Prequalification & Lead Scoring
- Use AI to prequalify borrowers early by scanning their preliminary documents (bank statements submission) and instantly estimate eligibility.
- Use AI models to score leads (likelihood to close) and prioritize outreach.
4. Borrower Interaction & Chatbots
- Embed AI chatbots that ask borrowers for required documents, guide them, and respond to FAQs (“how many months, what counts as deposit?”).
- Use NLP to interpret borrower emails or messages and trigger system actions (upload reminders, status updates).
5. Risk Monitoring & Portfolio Analytics
- Monitor live portfolio using AI to detect early delinquency patterns (e.g. drop in deposits or income).
- Build predictive models for default, rank portfolio risk, and feed insights to pricing / capital decisions.
Practical Implementation Roadmap
Here’s an example phased roadmap for launching a Non-QM Bank Statement product with AI in your organization:
| Phase | Objective | Key Actions |
|---|---|---|
| Phase 1: Product Definition & Underwriting Rules | Define product, rules, and fields | Finalize documentation requirements, thresholds, pricing spreads, exceptions |
| Phase 2: LOS Setup (Encompass) | Configure product within Encompass | Create product, custom fields, workflows, document groups |
| Phase 3: AI Tool Integration | Connect AI parsing, statement engine, underwriting engine | Integrate APIs, map outputs to custom fields, validate accuracy |
| Phase 4: Pilot & Testing | Test with a small batch of loans | Perform QA, compare AI results vs human, refine rules and thresholds |
| Phase 5: Training & Launch | Train teams, launch product | Roll out to originators, underwriters, monitor real-life metrics |
| Phase 6: Monitoring & Iteration | Track performance, improve models | Use closed loan results to retrain AI, adjust rule thresholds, tighten fraud detection |
Challenges & Risks – Mitigation
- Data quality & statement formats
Bank statements vary in format (banks, countries). Using robust OCR and normalization is critical. - Explainability & compliance
AI decisions (income parsing, risk scoring) must be explainable and auditable for underwriters and regulators. - Edge cases & exceptions
Unusual transactions, seasonality, or zero income months must be handled via overrides or manual review. - Model bias & overfitting
Ensure AI models aren’t biased against certain demographics; use caution and regular model validation. - Security & data privacy
Bank statement data is sensitive — compliance with encryption, storage, and privacy is mandatory. - Investor acceptance
Ensure that your product and AI methods are acceptable to investors or securitization investors.
Conclusion
Non-QM Bank Statement mortgage products unlock financing for borrowers who fall outside traditional underwriting constraints. Implementing such a product requires carefully designing documentation rules, underwriting logic, and LOS workflows (e.g. in Encompass). When augmented by AI — for statement parsing, income modeling, anomaly detection, and decision support — the process becomes more scalable, accurate, and efficient.
By combining human judgment and AI assistance, lenders can better serve underserved borrowers while controlling risk and maintaining competitiveness in the evolving mortgage marketplace.
