> For the complete documentation index, see [llms.txt](https://docs.3jane.xyz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.3jane.xyz/backing/fcc/credit-enhancement-and-loss-distribution.md).

# Credit Enhancement & Loss Distribution

This page maps the protection layers beneath USD3 and the loss distribution they sit on. Warehouse and forward-flow programs have had little prior history in DeFi, so their risks have rarely been mapped out in this context — this is that map.

## Layers of credit enhancement

Cash flowing to USD3 sits behind a stack of protection that absorbs losses in order:

1. **Pool excess spread** — the net yield each vintage generates over its life, before any principal is impaired.
2. **Originator first-loss equity** — the originator's own capital beneath 3Jane (e.g. a 25% first-loss slice → 1.33x overcollateralization).
3. **Overcollateralization (OC)** — the eligible borrowing base exceeds the drawn balance; tested on a schedule (e.g. weekly).
4. **Reserves & performance triggers** — cash reserves and covenant triggers that trap cash or accelerate amortization if performance deteriorates.
5. **sUSD3 subordination** — the junior tranche absorbs losses before USD3 at the 3Jane capital-stack level.

## Why granularity matters

<figure><img src="/files/icKPeLpo2mFd5RcRgHrK" alt="Loss distribution: single-name vs granular pool"><figcaption></figcaption></figure>

* Single-name corporate credit is **bimodal**: most loans pay at par, but a default jumps to restructuring-level loss severity.
* A granular SMB / consumer pool of \~3,000 obligors is **tight around expected loss**, because each obligor can fail independently.
* Expected loss can be similar across the two; the *shape* — and therefore the risk we are paid for — is completely different.

<figure><img src="/files/yYVIw6tkkL9olxye4AY3" alt="N=30 vs N=3,000 loss distributions"><figcaption><p>Same expected loss, very different distributions. With 30 obligors, unexpected loss is wide and fat-tailed; with 3,000 obligors the distribution collapses to a tight spike and standard deviation falls by roughly two orders of magnitude.</p></figcaption></figure>

At a 5% annual default probability and 50% LGD, expected pool loss is 2.5%. But the standard deviation of pool loss falls from \~2 percentage points at N=30 to \~0.2 percentage points at N=3,000. Diversification compresses idiosyncratic risk first; correlation risk is then handled at the structuring level.

<figure><img src="/files/UXxpDrGYKJgzHO0wiAPY" alt="OnDeck and Affirm ABS comparables"><figcaption><p>This is how the public ABS market routinely rates granular SMB and consumer receivables pools to investment grade. <em>Sources: OnDeck Asset Securitization Trust IV, Series 2023-1 (KBRA, July 2023); Affirm Asset Securitization Trust 2024-B (DBRS Morningstar, September 2024).</em></p></figcaption></figure>

## How structuring compresses the tail

<figure><img src="/files/oCsCkhStiK45JbfrI9y6" alt="Per-vintage capital stack"><figcaption><p>Capital stack on a per-vintage basis. Each cohort's net yield absorbs losses before any principal is impaired; the junior tranche absorbs anything beyond that; the senior takes losses only after both layers are exhausted. Markers show realised pool loss, the worst vintage in the book, and the senior break point.</p></figcaption></figure>

On the receivables book 3Jane underwrites, cumulative charge-offs sit around **1%** of disbursed principal. The worst single vintage came in around **4.5%**, and seasoned vintages collectively run under 2%. A facility example with \~4% of yield cushion and a 15% junior tranche puts USD3 / senior first-dollar principal loss at roughly **19% cumulative pool loss per vintage**.

## Correlation: borrowers failing together

<figure><img src="/files/5vZgT9OwXMGBVIcJiwMe" alt="Correlation stress test"><figcaption></figcaption></figure>

To model correlation risk, the structure was pressure-tested with a single-factor t-copula (ν = 10, fatter-tailed than Basel's standard Gaussian framework) at three intra-pool correlation regimes: ρ = 5% (benign), ρ = 15% (Basel's SMB base case), and ρ = 30% (GFC-equivalent correlated stress).

The senior tranche holds up across the relevant range. The 99th-percentile pool loss — a once-a-century outcome — sits at **11.0%** (benign) and **14.5%** (moderate), well within the junior tranche. Even at GFC-equivalent correlation it reaches just **19.9%**, right at the senior break. The senior takes meaningful losses only in the 1-in-1,000 tail combined with GFC-equivalent correlated stress.

<figure><img src="/files/eTiyrSm0AAjzh0Og5GGK" alt="Pool loss to tranche P&#x26;L"><figcaption><p>Propagation from pool loss to tranche P&#x26;L, scenario by scenario. The senior is untouched until pool losses on a vintage cross ~19%; the junior is paid for absorbing everything in between.</p></figcaption></figure>

{% hint style="warning" %}
Target yields and loss figures are illustrative and based on the historically observed performance of the receivables book 3Jane underwrites. They are not guarantees — realised losses, correlation, and recoveries can differ. See [Risks](/risks.md) for protocol-wide risk factors.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.3jane.xyz/backing/fcc/credit-enhancement-and-loss-distribution.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.