Individual credit limits are calculated primarily through an expected loss framework that combines probability of default, loss given default, and exposure at default, supplemented by potential future exposure modeling for derivatives and trading lines. Getting these parameters right matters more than ever, given the current market backdrop: Fitch reported a 5.8% default rate for US private credit in the twelve months through January 2026, the highest on record, in a segment where borrowers typically carry no public rating at all.<\/p>\n\n\n\n
That combination, rising defaults and thinning external reference points, is exactly why the methodology behind an individual limit deserves scrutiny rather than a template approach. Here is how institutions actually build that number.<\/p>\n\n\n\n
The core calculation underlying most individual credit limits is the expected loss formula: EL equals PD multiplied by LGD multiplied by EAD. Each component answers a distinct question.<\/p>\n\n\n\n
Probability of default estimates the likelihood a specific counterparty fails to meet its obligations within a given time horizon. Loss given default estimates what percentage of the exposure would actually be lost if that default occurred, after accounting for recoveries from collateral, guarantees, or bankruptcy proceedings. Exposure at default estimates the total amount outstanding at the moment of default, including principal, accrued interest, and any undrawn commitments that could be drawn down as a borrower approaches distress.<\/p>\n\n\n\n
Multiplying these three figures together produces the expected loss on a specific exposure, which then feeds directly into how large a limit the institution can responsibly extend to that counterparty.<\/p>\n\n\n\n
Loan exposure is relatively static once drawn, but derivative exposure moves with the market every day, which requires a different methodology. Institutions model potential future exposure, often at a 95th percentile confidence level, to capture the worst-case exposure they should prepare for when setting a limit.<\/p>\n\n\n\n
A worked example illustrates the mechanics. Consider a five-year interest rate swap on a $50 million notional. At its peak, expected exposure might run around $1.8 million, the average amount a counterparty would owe across simulated market scenarios. The 95th percentile PFE, however, might reach $3.2 million, representing the exposure level the limit actually needs to account for. If that counterparty carries a 2% annual probability of default and a 60% loss given default, the expected loss at that peak point comes to roughly $21,600, a figure regulators and internal risk teams both use to size the limit appropriately.<\/p>\n\n\n\n
EAD calculation methodology varies significantly by product type, and applying the wrong approach understates real exposure.<\/p>\n\n\n\n
Under an ISDA Master Agreement, for example, all derivative trades between two counterparties get treated as a single net obligation upon default, which materially changes the exposure figure feeding into the limit calculation compared to treating each trade in isolation.<\/p>\n\n\n\n
Alongside the quantitative expected loss framework, institutions widely use rating-based approaches, where a counterparty’s internal or external credit rating maps directly to a maximum limit band. A stronger rating supports a higher ceiling, while a weaker rating triggers a lower cap or additional collateral requirements.<\/p>\n\n\n\n
This approach works cleanly for rated corporates and sovereigns but runs into a structural problem for a growing share of today’s lending activity, which brings us to the harder case.<\/p>\n\n\n\n
Once PD, LGD, and EAD are established for a counterparty, the actual limit typically reflects a maximum acceptable expected loss or a maximum acceptable capital consumption under the institution’s Basel IRB framework, where the regulatory capital charge is sized specifically against unexpected loss. If the PD feeding that calculation runs too low, the institution ends up holding less capital than its actual risk requires, which is why model accuracy carries direct regulatory consequences, not just internal risk management ones.<\/p>\n\n\n\n
Direct-lending funds, middle-market companies, private equity portfolio companies, and fund counterparties driving much of today’s private credit growth almost never carry a public rating. The Financial Stability Board noted in May 2026 that this makes it hard to monitor risk across the market, particularly since these borrowers tend to carry lower credit quality and higher leverage than comparable, observable borrowers.<\/p>\n\n\n\n
For these counterparties, institutions increasingly rely on consensus data, pooling risk views from other lenders with exposure to the same borrower, since it is often the only external reference point available when no market price or rating exists. Reduced-form, or hazard-rate, models have become the standard tool for building PD term structures across multiple horizons in these cases.<\/p>\n\n\n\n
Data scarcity for unrated counterparties<\/strong> Internal models can be well-built and well-documented and still lack the independent benchmark needed to confirm they are calibrated against reality, particularly for private and cross-border borrowers.<\/p>\n\n\n\n Parameters that move together<\/strong> PD, LGD, and EAD are not independent in practice. When defaults spike, collateral values often fall, linking LGD to PD, and when credit quality deteriorates, borrowers tend to draw more of their available credit, linking EAD to PD as well.<\/p>\n\n\n\n Regulatory scrutiny on model inputs<\/strong> IRB models face deeper and more frequent supervisory scrutiny around statistical robustness and representativeness, which raises the cost of a poorly calibrated PD or LGD estimate feeding into a limit.<\/p>\n\n\n\nBest Practices for Calculating Individual Credit Limits<\/h3>\n\n\n\n