> 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 {% hint style="info" %} All numbers below are illustrative of the underlying structure and are not a reflection of the current USD3 backing. {% endhint %} This page maps the protection layers beneath each tranche 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 Each tranche has its own stack of protection. ### USD3 credit enhancement Cash flowing to USD3 sits behind a stack of protection that absorbs losses in order: 1. **Originator first-loss equity** — the originator's own capital beneath 3Jane (e.g. a 25% first-loss slice → 1.33x overcollateralization). 2. **Overcollateralization (OC)** — the eligible borrowing base exceeds the drawn balance; tested on a schedule (e.g. weekly). 3. **Reserves & performance triggers** — cash reserves and covenant triggers that trap cash or accelerate amortization if performance deteriorates. 4. **sUSD3 subordination** — the junior tranche absorbs losses before USD3 at the 3Jane capital-stack level. ### sUSD3 credit enhancement sUSD3 absorbs losses ahead of USD3, but it is not first-dollar-loss capital. Its enhancement layer is **pool excess spread** — the net yield the pool generates, which absorbs losses before any principal is impaired. * In a traditional deal the junior owes the senior a fixed coupon and eats any yield shortfall — it can lose money with zero defaults. * 3Jane has no fixed coupon: both tranches earn a proportional share of pool yield (see [Pool Interest Rates](/usd3-susd3/pool-interest-rates.md)), so there is no shortfall to absorb. * Losses net against pool yield before touching principal; sUSD3 principal is impaired only when realized losses exceed that yield. * USD3 stands behind sUSD3, so this yield buffer is also the first layer in the overall loss order. ## Why granularity matters
Loss distribution: single-name vs granular pool
* 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.
N=30 vs N=3,000 loss distributions

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.

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.
OnDeck and Affirm ABS comparables

This is how the public ABS market routinely rates granular SMB and consumer receivables pools to investment grade. Sources: OnDeck Asset Securitization Trust IV, Series 2023-1 (KBRA, July 2023); Affirm Asset Securitization Trust 2024-B (DBRS Morningstar, September 2024).

## How structuring compresses the tail
Per-vintage capital stack

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.

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
Correlation stress test
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.
Pool loss to tranche P&L

Propagation from pool loss to tranche P&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.

{% 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 %}