Collateral analysis

Every ABF deal starts with a loan tape. Before you discuss advance rates, negotiate triggers, or build cash flow models, you need to understand what’s actually in the pool. Collateral analysis answers the foundational question: what are these assets, how are they performing, and what do the patterns tell you about future behavior?

This topic covers the mechanics of tape analysis, from initial data validation through static pool construction and performance metric calculation. Whether you’re an originator preparing for a capital raise or a capital provider evaluating a new opportunity, these are the techniques you’ll use daily.

Subtopics

Loan tape fundamentals

What a loan tape is, standard fields across asset classes, the data dictionary, format issues to watch for, and the first five checks to run when a new tape arrives.

Data validation and quality

Field-level validation, logical consistency checks, duplicate detection, exception reporting, common data quality issues, and how to address problems you find.

Stratification analysis

Essential stratifications (balance, rate, credit, geography, vintage), reading strat tables for patterns, concentration analysis, HHI calculation, and excess concentration haircuts.

Static pool analysis

Building static pool data, cumulative loss curves, vintage comparison, seasoning adjustments, and loss development methods for projecting newer vintages.

Performance metrics

CDR, CPR, CNL, and severity calculation and interpretation. Roll rate analysis and transition matrices. Forward default projection and benchmarking.

The analysis workflow

Collateral analysis follows a systematic workflow:

Receive and validate tape — Confirm data completeness, run field-level checks, document exceptions
Run stratifications — Build the standard strat package to understand pool composition
Identify concentrations — Check against limits, calculate HHI, size any required haircuts
Analyze static pools — Track vintage performance, build loss curves, project terminal losses
Calculate performance metrics — CDR, CPR, severity, roll rates for the current snapshot
Benchmark — Compare to historical performance and comparable public deals
Present findings — Executive summary, formatted tables, trend analysis, red flag documentation

Presenting results

Your collateral analysis isn’t useful if it’s buried in spreadsheets. Investment committee members and lenders need to absorb your findings quickly and understand the implications.

The executive summary

Lead with the answer, not the methodology:

Collateral summary: XYZ Consumer Pool

The $50M pool contains 4,000 loans with a weighted average FICO of 698 and WAC of 10.3%. Current 30+ DQ is 7.0%, elevated versus the 5% benchmark for near-prime consumer. California concentration is 24%, exceeding the 20% limit by $2M.

Key findings:

Data quality is adequate (98% field completion, 12 exceptions documented)

Credit quality is consistent with stated guidelines (no material outliers)

Q4 2023 vintage is underperforming Q1-Q3 by approximately 80 bps at equivalent age

Concentration issue requires either limit exception or $1M borrowing base reduction

Recommendation: Proceed to full diligence with concentration haircut applied.

Formatted strat tables

Present strats with clear headers, consistent formatting, and relevant weighted averages:

Credit band	Count	% Count	UPB ($M)	% UPB	Avg rate	Avg balance
750+	600	15.0%	$10.5	21.0%	8.2%	$17,500
700-749	1,200	30.0%	$17.5	35.0%	9.5%	$14,583
660-699	1,400	35.0%	$15.0	30.0%	11.2%	$10,714
620-659	600	15.0%	$5.5	11.0%	13.5%	$9,167
<620	200	5.0%	$1.5	3.0%	15.8%	$7,500
Total/WA	4,000	100%	$50.0	100%	10.3%	$12,500

Illustrative pricing. See pricing disclaimer.

Add a line of commentary below each strat: “Pool skews near-prime; 3% subprime tail is within guidelines. Credit-rate alignment is appropriate.”

Trend analysis

Compare the current tape to prior periods:

Metric	Current tape	Prior tape (3mo ago)	Change
Total UPB	$50.0M	$42.5M	+17.6%
Loan count	4,000	3,400	+17.6%
WA FICO	698	702	-4 pts
WAC	10.3%	10.0%	+30 bps
30+ DQ	7.0%	5.8%	+120 bps

Commentary: “Pool growth is consistent with facility ramp. FICO decline of 4 points and DQ increase of 120 bps warrant monitoring but remain within guidelines. Higher WAC suggests appropriate risk-adjusted pricing.”

Red flag documentation

Be explicit about concerns:

Red flags identified:

California concentration (24% vs. 20% limit): Requires either limit waiver or $2M exclusion from eligible pool

Q4 2023 vintage performance (CNL tracking 80 bps above prior vintages at equal age): Recommend vintage-level eligibility review

Missing FICO (127 loans, $1.55M): Originator to confirm these loans meet alternative credit criteria per guidelines

Don’t bury concerns in footnotes. If something would change your credit decision or pricing, highlight it.

Tools

What you can do in Excel

For pools under 50,000 loans, Excel handles most analysis:

Pivot tables for stratifications
SUMIFS/COUNTIFS for custom cuts
Chart tools for static pool visualization
Basic data validation

Limitations: Slow on large files, no version control, formula errors are easy to miss.

When to move to code

Move to Python/SQL when:

Loan count exceeds 100,000
Analysis needs to be repeatable monthly
Complex calculations (roll rates, transition matrices)
Data from multiple sources needs joining

Python libraries: pandas for data manipulation, matplotlib/seaborn for visualization, numpy for calculations.

SQL: Essential if tape data lives in a database. Most large originators provide tape access via SQL query rather than file export.

Commercial platforms

Platform	Best for
Intex	Rated ABS cash flows, deal structuring
Bloomberg	Market data, pricing, surveillance
Moody’s Analytics	Credit modeling, stress testing
Trepp	CMBS/CLO specific analytics

For warehouse facilities and private deals, you often don’t need these. They’re essential for rated transactions and secondary trading.

Where to find benchmark data

Your analysis needs context. A 5% CDR means nothing without knowing whether that’s good, bad, or typical for the asset class.

Public ABS performance

EDGAR ABS-EE filings: SEC-required loan-level data for registered ABS. Contains monthly performance data including DQ, losses, and prepayments. Free at sec.gov/edgar.

10-D filings: Monthly distribution reports for public ABS. Include pool-level performance summaries, waterfall calculations, and trigger status.

Rating agency data

S&P Global Ratings: Transition and default studies by asset class. Annual performance reports for auto, consumer, and other ABS.

Moody’s Investors Service: Asset-level data and performance indices. Moody’s Analytics STAR database for auto loan performance.

Fitch Ratings: Consumer ABS index, auto ABS indices, periodic surveillance reports.

Most reports are available free on rating agency websites under “Research” or “Publications.”

Using benchmarks

Don’t use benchmarks blindly. Adjust for:

Credit quality: Benchmark pool prime vs. your pool near-prime = expect 100-200 bps higher losses
Vintage: 2023 originations vs. 2019 originations perform differently
Product type: Secured vs. unsecured, term vs. revolving
Structure: Public ABS (tight eligibility) vs. warehouse (broader eligibility)

Build a comps table showing your pool’s characteristics alongside 3-5 comparable public deals. Explain the adjustments you’re making.

Practical checklist

Before declaring your collateral analysis complete:

Key takeaways

Start with validation. Data completeness and quality checks take an hour and prevent building analysis on corrupted data.
Strats are your primary tool. Run balance, rate, credit, geography, and vintage strats on every pool. Look for skew, outliers, and concentration.
Static pool analysis reveals truth. Dynamic portfolio metrics are diluted by new originations. Static pools show actual performance by vintage.
Benchmark everything. A 5% CDR means nothing without context. Find comparable deals and adjust for differences.
Present findings clearly. Lead with the executive summary. Highlight red flags. Don’t bury concerns in footnotes.