Capital One (COF) — Hard 10% APR Cap Stress Analysis
Credit Card Unit Economics Assessment | Q3 2025
1. EXECUTIVE SUMMARY
Verdict: SHORT
Policy Beta: HIGH
Key Conclusions (8 bullets max):
- Model shows unprofitability at 10% APR: At a hard 10% cap, quarterly segment pre-tax economics swing from +$3.1B to -$2.3B (per APR_Sensitivity!B2), a ~$5.4B quarterly reversal on the Credit Card segment alone.
- Breakeven APR sits at 13.4% (3Q25 annualized, Summary!B5), meaning even a "soft" 12.5% cap would leave the card business loss-making without significant de-risking.
- Interest income dependency is extreme: 84.6% of gross card revenue derives from interest income (Scoring_Inputs!C3), leaving minimal fee/interchange buffer to absorb yield compression.
- Subprime exposure manageable but non-trivial: 27% of loans carry credit scores ≤660 (Inputs!B16), representing the cohort most likely rationed under cap scenarios—expect meaningful balance shrinkage if underwriting tightens.
- Credit metrics currently improving but won't save profitability: 3Q25 NCO rate of 4.61% (Mini_PnL!B19) is down 99bps YoY, yet net losses alone consume >45% of current interest income—under a 10% cap, losses would exceed total yield.
- Diversification insufficient to protect consolidated earnings: Credit Card represents 75.6% of company-wide net revenue (Scoring_Inputs!C13); no realistic offset from Consumer Banking or Commercial Banking.
- Discover integration complicates mitigation timing: Management signaled a "brownout" in Discover portfolio growth for "the next couple of years" (Transcript, p.7), constraining near-term balance expansion even if caps are avoided.
- Capital return program signals confidence but increases downside risk: $16B buyback authorization (Transcript, p.3) at 14.4% CET1 is aggressive; a severe cap scenario could pressure both earnings power and capital buffers simultaneously.
2. WHAT THE MODEL SAYS (Mechanical Outputs)
A. Base Unit Economics — Per $100 Average Loans (Annualized)
| Line Item | 3Q25 | LTM (4Q24–3Q25) | Source |
|---|
| Gross interest income | $17.99 | $18.30 | Mini_PnL!D13 |
| Non-interest income | $3.29 | $3.59 | Mini_PnL!D14 |
| Gross revenue | $21.28 | $21.88 | Mini_PnL!D15 |
| Less: Funding cost (implied) | ($4.03) | ($4.10) | Mini_PnL!D16 |
| Net interest income | $13.96 | $14.19 | Mini_PnL!D17 |
| Less: Opex | ($8.04) | ($8.75) | Mini_PnL!D18 |
| Less: Credit losses (NCO) | ($4.61) | ($5.35) | Mini_PnL!D19 |
| Economic pre-tax profit | $4.60 | $3.68 | Mini_PnL!D20 |
Interpretation: Current economics generate ~$4.60 per $100 of balances annually. Breakeven requires yield >13.4% (Summary!B5).
B. APR Sensitivity — Economic Pre-Tax at Various Yield Levels
| Assumed APR/Yield | Interest Income ($mm, qtr) | Gross Revenue ($mm) | Econ Pre-Tax ($mm) | Pre-Tax per $100 | Profitable? |
|---|
| 10.0% | 6,729 | 8,940 | (2,283) | (3.39) | No |
| 12.5% | 8,412 | 10,623 | (600) | (0.89) | No |
| 15.0% | 10,094 | 12,305 | 1,082 | 1.61 | Yes |
| 17.5% | 11,776 | 13,987 | 2,764 | 4.11 | Yes |
| 17.99% (Current) | 12,109 | 14,320 | 3,097 | 4.60 | Yes |
Source: APR_Sensitivity!B2:G7
Key Finding: At 10% yield, the card segment generates a quarterly loss of $2.28 billion before any de-risking response.
C. Breakeven APR Calculation
| Metric | 3Q25 | LTM |
|---|
| Breakeven APR (economic) | 13.39% | 14.62% |
Source: Mini_PnL!B21, Summary!B5
Drivers of breakeven:
- Funding cost: 4.03% (implied, Inputs!B17)
- Opex intensity: 8.04 per $100 (Mini_PnL!D18)
- NCO rate: 4.61% annualized (Inputs!B11)
- Non-interest income offset: 3.29 per $100 (Mini_PnL!D14)
D. Credit Quality Metrics
| Metric | Value | Period | Source |
|---|
| Net charge-off rate | 4.61% | 3Q25 ann. | Inputs!B11 |
| 30+ day performing delinquency | 3.84% | Sep 30, 2025 | Inputs!B12 |
| 90+ day delinquent still accruing | 0.01% | Sep 30, 2025 | Inputs!B13 |
| Allowance coverage ratio | 7.28% | Sep 30, 2025 | Inputs!B14 |
| Credit score mix >660 | 73% | Sep 30, 2025 | Inputs!B15 |
| Credit score mix ≤660 | 27% | Sep 30, 2025 | Inputs!B16 |
E. De-Risking Sensitivity (Judgment Overlay)
The Excel does not model explicit de-risking scenarios, but we can construct them using disclosed sensitivities:
Assumptions for de-risking under 10% cap:
- Balance shrinkage: 15–30% (rationing high-yield revolvers)
- NCO improvement: 100–150bps (better credit mix post-cut)
- Funding cost benefit: 0–25bps (lower balance = less wholesale funding)
| Scenario | Balance Reduction | NCO Rate | Funding Cost | Implied Econ Pre-Tax (qtr) |
|---|
| No de-risking | 0% | 4.61% | 4.03% | (2,283) |
| Mild de-risking | 15% | 3.61% | 3.90% | (1,200) to (1,500) |
| Aggressive de-risking | 30% | 3.11% | 3.75% | (600) to (900) |
These are JUDGMENT overlays, not Excel-derived outputs.
F. Funding Cost Proxies
| Metric | Rate | Source |
|---|
| Implied card segment funding cost | 4.03% | Derived: Inputs!B17 ÷ Inputs!B2 |
| Interest-bearing deposit rate (consolidated) | 3.27% | Inputs!B18 |
| Total interest-bearing liabilities cost | 3.55% | Inputs!B19 |
| Non-interest-bearing deposits share | 5.9% | Scoring_Inputs!C10 |
Note: The segment does not disclose explicit FTP allocation. The 4.03% implied rate is derived from gross interest income less segment net interest income, divided by average loans.
G. Key Limitations/Flags (from Missing_Data sheet)
| Missing Item | Impact on Analysis | Workaround |
|---|
| FICO 740+/660–739/<660 bands | Cannot separate prime vs. superprime exposure | Used disclosed 660 threshold only |
| Average portfolio FICO | Lower confidence in mix attribution | Used delinquency/NCO trends as signal |
| Rewards expense (separate) | Cannot model reward-cut lever | Embedded in segment economics |
| Card segment FTP/funding cost | Breakeven APR dependent on implied allocation | Anchored to disclosed totals |
| Interchange vs. fee split | Cannot separately stress interchange regulation | Used total non-interest income |
| Opex split (marketing/servicing/tech) | Cannot identify variable costs | Used total segment non-interest expense |
Source: Missing_Data sheet
3. EARNINGS-AT-RISK (12-Month Range)
A. Mechanical Calculation
Step 1: Delta to segment pre-tax at 10% yield
| Metric | Current (17.99%) | At 10% Cap | Delta |
|---|
| Quarterly econ pre-tax ($mm) | 3,097 | (2,283) | (5,380) |
| Annualized segment impact ($mm) | 12,388 | (9,132) | (21,520) |
Source: APR_Sensitivity!B2, B7
Step 2: Convert to dollars using average receivables
Average loans HFI (3Q25): $269,175mm (Inputs!B2)
At 10% yield:
- Annual interest income: $269,175 × 10% = $26,918mm (vs. $48,436mm at 17.99%)
- Interest income haircut: ~$21.5B annualized
Step 3: Apply de-risking overlay (range)
| Scenario | Balance Shrink | NCO Improvement | 12-Mo Segment Pre-Tax Impact |
|---|
| Low (aggressive mitigation) | 30% | 150bps | (7.0)B to (9.0)B |
| Base (moderate mitigation) | 20% | 100bps | (9.0)B to (12.0)B |
| High (limited mitigation) | 10% | 50bps | (15.0)B to (18.0)B |
JUDGMENT overlay — not Excel-derived
Step 4: Company-level translation
Credit Card segment contribution to total company net revenue: 75.6% (Scoring_Inputs!C13)
Segment pre-tax as % of consolidated pre-tax (3Q25): ~87% ($3,834mm / $4,382mm, 8-K Supplement Table 9)
Estimated consolidated pre-tax impact range (12 months):
- Low: (6)B to (8)B
- Base: (8)B to (11)B
- High: (13)B to (16)B
For context: 3Q25 consolidated GAAP pre-tax income was $4.4B ($17.5B annualized). A base-case 10% cap scenario would wipe out ~50–65% of normalized earnings.
4. QUALITATIVE SCORECARD + COMPOSITES
A. Dimension Scores (0–10, higher = better positioned)
| Dimension | Score | Confidence | Rationale |
|---|
| Revolver intensity | 2 | High | 84.6% interest share of gross revenue (Scoring_Inputs!C3); yield of 17.99% implies heavy revolver mix; among highest in industry. |
| Credit spectrum / rationing risk | 5 | Medium | 27% ≤660 exposure (Inputs!B16) is manageable but represents first-loss cohort under cap; absence of FICO granularity limits precision. |
| Funding advantage | 6 | Medium | 3.27% deposit cost (Inputs!B18) is competitive; 5.9% non-interest-bearing share (Scoring_Inputs!C10) provides modest buffer but less than deposit-heavy peers. |
| Mitigation levers | 3 | Medium | Fee share only 15.4% (Scoring_Inputs!C11); rewards expense unknown; opex intensity 46.6% (Scoring_Inputs!C12) suggests limited quick-win cuts; underwriting tightening feasible but destroys balances. |
| Diversification / offset capacity | 3 | High | Credit Card is 75.6% of net revenue (Scoring_Inputs!C13); Consumer Banking and Commercial Banking cannot absorb a card segment collapse; no realistic 12-month offset. |
B. Composite Scores
Earnings-at-Risk Score (0–100; higher = WORSE exposure)
| Component | Weight | Raw Score | Weighted |
|---|
| Revolver intensity | 35% | 8 | 2.80 |
| Credit spectrum risk | 25% | 5 | 1.25 |
| Funding disadvantage | 15% | 4 | 0.60 |
| Diversification gap | 25% | 7 | 1.75 |
| Total | 100% | | 64 |
Interpretation: 64/100 — High earnings-at-risk under a hard cap scenario.
Survivability Score (0–100; higher = BETTER ability to sustain acceptable ROE)
| Component | Weight | Raw Score | Weighted |
|---|
| Mitigation levers | 30% | 3 | 0.90 |
| Funding advantage | 20% | 6 | 1.20 |
| Credit quality trajectory | 20% | 6 | 1.20 |
| Capital cushion (14.4% CET1) | 15% | 7 | 1.05 |
| Management optionality | 15% | 4 | 0.60 |
| Total | 100% | | 50 |
Interpretation: 50/100 — Marginal survivability; firm would survive but ROE would collapse to low single digits or worse for multiple years.
5. 12-MONTH OPERATIONAL NARRATIVE
What Actually Happens Under a Hard 10% Cap
Quarter 1–2 (Announcement + Positioning):
- Management immediately halts new originations to high-balance revolvers and deep subprime cohorts.
- Marketing spend ($1.4B/quarter currently, Transcript p.2) gets slashed 30–40% as customer acquisition ROI inverts.
- Rewards programs face renegotiation pressure, but redemption liabilities limit rapid cuts.
- Balance runoff begins as maturing loans exceed constrained new originations.
- Credit spreads on COF unsecured debt widen 100–150bps as rating agencies issue negative outlooks.
Quarter 3–4 (Structural Adjustment):
- Portfolio shrinks 10–20% as high-yield accounts attrit or are proactively closed.
- NCO rates improve 50–100bps as subprime exits the book, but dollar losses remain elevated due to back-book seasoning.
- Opex rationalization begins (headcount, servicing costs), but integration with Discover constrains flexibility.
- Interchange income declines with purchase volume as rationed customers reduce spending.
- Capital position tested: if earnings collapse, buyback program pauses; dividend sustainability questioned.
Who Loses Access to Credit:
- Deep subprime (<620 FICO): Effectively exits the credit card market entirely.
- Subprime (620–660): Severe line reductions; new approvals near zero.
- Near-prime (660–700): Modest line cuts; higher qualification hurdles.
- Prime/superprime: Minimal impact; may see reduced promotional offers.
Second-Order Effects:
- Volume: Purchase volume growth (currently +6.5% ex-Discover, Transcript p.4) reverses to -5% to -10% as rationed customers migrate to debit.
- Rewards: Pressure to cut rewards erodes brand equity in premium segment (VentureX); competitors with higher fee models gain share.
- Losses: Near-term NCO spike possible as "exit cohorts" charge off; then structural improvement.
- Funding: Deposit franchise (Discover synergy) provides stability, but ABS markets may re-price spread.
- Discover integration: Synergy timeline extends as focus shifts to survival; $2.5B synergy target at risk.
6. VERDICT + "CHANGE MY MIND" TRIGGERS
Verdict: SHORT
3 Key Reasons (Grounded in Model Outputs):
- Unprofitable under cap: Economic pre-tax swings from +$3.1B to -$2.3B quarterly at 10% yield (APR_Sensitivity!B2, B7). No realistic combination of fee increases, cost cuts, or de-risking restores profitability within 12 months.
- Limited mitigation levers: Fee share of 15.4% (Scoring_Inputs!C11) provides negligible offset; opex intensity of 46.6% (Scoring_Inputs!C12) is already lean by card standards; rewards cuts damage competitive position.
- Concentration risk: 75.6% net revenue dependence on Credit Card (Scoring_Inputs!C13) means no diversification buffer; Consumer Banking and Commercial cannot absorb the shock.
"Change My Mind" Triggers (3–5 Specific):
- Regulatory clarity on exemptions: If any cap legislation includes exemptions for rewards cards, secured cards, or balance-transfer products, the exposed book shrinks materially. Watch for: Legislative text, CFPB guidance, bank lobby amendments.
- Faster-than-expected NCO improvement: If quarterly NCO rate drops below 3.5% (vs. 4.61% current), breakeven APR falls toward 11–12%, narrowing the shortfall. Watch for: DQ1 trends in 4Q25/1Q26, recovery rates, vintage performance.
- Aggressive fee restructuring: If management announces annual fee increases or new fee products that lift non-interest income share above 25% of gross revenue (vs. 15.4% today), the economics improve. Watch for: Product repricing announcements, fee income disclosures.
- Material balance shrinkage with stable revenue: If management can shrink the portfolio 20–30% while maintaining 70–80% of revenue (via pricing/mix), survivability improves. Watch for: Loan growth guidance, attrition commentary, NIM disclosure.
- Discover network monetization acceleration: If debit migration to Discover network generates >$2B incremental annual revenue (vs. embedded synergy assumptions), this provides partial offset. Watch for: Network volume disclosures, interchange rate commentary.
Policy Probability Caveat (5 bullets):
- Treat probability as similar across major card issuers: All large issuers (AXP, DFS, SYF, COF) face the same legislative uncertainty; relative positioning, not absolute probability, drives trade selection.
- Hard 10% cap is a tail-risk scenario, not base case: Most legislative proposals have stalled; political realities favor compromise (e.g., 15–18% caps with exemptions). Stress testing the hard case maps the downside, not the expected path.
- Usefulness of downside mapping: Even if probability is 10–20%, the severity of impact justifies premium paid for downside protection. Asymmetric payoff profile favors hedged positions.
- Scenario value is in relative ranking: COF ranks among the most exposed large-cap banks due to revolver intensity and concentration; this relative weakness persists across cap levels (10%, 12.5%, 15%).
- Timing uncertainty is high: Legislative timelines are unpredictable; the trade requires patience and/or option-based expression to manage carry.
7. MISSING DATA AND CONFIDENCE
Top Missing Items and Sharpening Effects
| Missing Item | Sharpening Effect |
|---|
| FICO distribution (740+/660–739/<660) | Would allow precise sizing of "at-risk" book under rationing; currently using 27% ≤660 as proxy, but cannot separate near-prime from deep subprime. |
| Rewards expense separately disclosed | Would clarify flexibility to cut rewards without P&L disclosure; current analysis assumes rewards are embedded in opex/non-interest income. |
| Card segment FTP/funding cost | Would improve breakeven APR precision; current 4.03% implied rate may include noise from other allocations. |
| Fee elasticity | Would inform mitigation lever sizing; no disclosed data on customer response to annual fee increases. |
| Revolver/transactor mix | Would directly quantify interest income dependency; yield of 17.99% implies heavy revolver mix, but exact split unknown. |
| APR distribution by credit tier | Would allow tier-specific impact modeling; currently treating portfolio as homogeneous yield. |
| Rewards liability duration | Would inform timing of potential cuts; breakage and redemption assumptions not disclosed. |
Confidence Assessment
| Analysis Component | Confidence | Rationale |
|---|
| Base unit economics | High | Fully derived from disclosed segment P&L and reconciled to 10-Q. |
| APR sensitivity mechanics | High | Direct calculation from interest income and cost structure. |
| De-risking overlays | Medium | JUDGMENT-based; no disclosed sensitivity. |
| Earnings-at-risk range | Medium | Anchored to Excel but requires assumptions on balance shrinkage and NCO improvement. |
| Survivability score | Low-Medium | Qualitative; depends on unobservable management actions and regulatory response. |
| Relative positioning | High | COF's revolver intensity and concentration are verifiable facts. |
Analysis prepared using Capital One Q3 2025 disclosures: 8-K Supplement, 10-Q, Earnings Presentation, Earnings Transcript, and pre-built Excel unit economics model. All quantitative outputs cite sheet!cell references. Judgment overlays explicitly labeled.
END OF REPORT