Underwriting High-Yield GCC Credit: Pricing the Risk Behind the Coupon

Introduction

The previous question, where the Gulf's double-digit coupons come from, has a structural answer: they are predominantly compensation for illiquidity, complexity and the supply of scarce capital, with a smaller credit-spread component for expected loss. That answer is reassuring in aggregate but useless on a single transaction, because it describes the market average rather than the deal in front of the lender. The coupon on a specific transaction may adequately price its specific risk, or it may not, and the difference between the two is the difference between a sound loan and a loss waiting to happen. This paper is about that difference. It asks not where the coupon comes from but whether, on any given deal, the coupon is enough.

The discipline the paper advocates inverts the instinct of yield-driven investing. A yield-driven lender starts with the coupon and asks whether it is attractive; an underwriting-driven lender starts with the risk and asks what coupon would be required to compensate for it, then compares that required coupon with the one on offer. If the offered coupon exceeds the required coupon, the transaction creates value; if it falls short, it destroys value however high it looks. The required coupon is the output of an underwriting process, and the quality of that process determines the quality of the lending book. The coupon on offer is a fact; the coupon required is a judgement, and the lender's edge lies in making that judgement well.

Two quantities govern the required coupon: the probability that the borrower defaults, and the fraction of the exposure the lender loses if it does. The product of the two is the expected loss, the through-the-cycle average cost of credit that the coupon must first cover before it earns anything. Beyond expected loss, the lender must hold capital against the possibility that losses exceed the average in a bad year, the unexpected loss, and must earn a return on that capital. The required coupon is therefore the sum of the funding cost, the expected loss, the cost of the capital held against unexpected loss, the operating cost of underwriting and servicing, and the target margin. This paper builds that sum component by component and shows how it varies across transactions.

The stakes are highest precisely where the coupon is highest. The transactions that quote twenty per cent are, by the logic of the companion paper, those with the largest credit-spread component, the most subordinated claims on the weakest collateral and sponsors. These are the transactions where underwriting matters most, because the coupon is genuinely compensation for a credit risk that must be estimated accurately, and where the cost of getting it wrong is greatest. A disciplined framework is not a bureaucratic overlay on these transactions; it is the difference between earning the credit spread and funding the loss it was meant to compensate.

Results And Discussion

This section builds the pricing framework from its components: the expected-loss matrix, the loss-given-default decomposition, the risk-based required-coupon curve, the underwriting scorecard, and the relationship between underwriting quality and realised loss.

The Expected-Loss Matrix

Figure 1 presents expected loss as a matrix over probability of default and loss given default. Reading the matrix makes the first discipline of underwriting concrete: expected loss is the product of two quantities, and a low value of either keeps the expected loss modest even when the other is high. A transaction with a high default probability but strong collateral, low loss given default, can have a lower expected loss than one with a moderate default probability but weak collateral. The matrix warns against assessing either dimension in isolation, and it locates the dangerous transactions in the bottom-right, high default probability and high loss given default together, where expected loss escalates fastest.

Figure 1. The Expected-Loss Matrix: PD times LGD

Expected loss is the product of default probability and loss given default; strong collateral can offset high default risk.

The matrix also frames the lender's task. Default probability is largely a property of the borrower and the environment, which the lender assesses but does not control; loss given default is largely a property of the collateral and structure, which the lender negotiates. Moving a transaction leftward in the matrix, reducing default probability, is mostly a matter of selection; moving it upward to a lower loss given default is a matter of structuring. The lender's leverage, per Proposition 2, lies disproportionately in the second.

Decomposing Loss Given Default

Figure 2 decomposes loss given default for a representative secured transaction, building from the gross collateral value down to the net recovery through the valuation haircut, the forced-sale discount, enforcement costs and the time discount. The decomposition reveals how a loan that looks comfortably over-collateralised on day one can deliver a disappointing recovery in default, because each deduction erodes the cushion, and the deductions compound. A nominal loan-to-value that looks safe can translate into a material loss once the haircuts, discounts, costs and delays of a real enforcement are applied.

Figure 2. Building Loss Given Default: From Collateral to Net Recovery

Each deduction, haircut, sale discount, enforcement cost and time, erodes the collateral cushion; they compound.

The practical lesson is that the headline loan-to-value overstates the protection. A lender that underwrites to the appraised value, without applying the haircuts and discounts that a real enforcement imposes, will systematically underestimate its loss given default and so under-price its credit. The disciplined lender underwrites to the net recovery, not the gross collateral, and structures the security to minimise each deduction: faster enforcement, clearer title, first-ranking charges and cash control all raise net recovery, and each is a lever the lender can pull at origination.

Recovery and Loan-to-Value

Figure 3 (presented after the recovery analysis) and Figure 7 together show how net recovery falls as loan-to-value rises. At low loan-to-value the collateral cushion absorbs the haircuts and discounts and recovery is near complete; as loan-to-value rises, the cushion thins and recovery falls, slowly at first and then sharply once the cushion is exhausted. The relationship is non-linear, which is why marginal increases in leverage at high loan-to-value are so dangerous: each additional point of leverage removes a point of cushion exactly where the cushion is most needed. The lender's loan-to-value discipline is, in effect, its loss-given-default discipline.

Figure 3. Net Recovery as a Function of Loan-to-Value

Recovery is near-complete at low leverage and falls sharply once the collateral cushion is exhausted.

The Risk-Based Required-Coupon Curve

Figure 8 presents the central output of the framework: the coupon a rational lender should require as a function of the probability of default, decomposed into the credit-spread component and the rest. The curve rises with default probability, as it must, because higher default risk demands higher compensation; and the credit-spread component, the part that compensates specifically for loss, rises faster than the total, because at higher default probabilities more of the coupon must go to covering and capitalising loss rather than to margin. The curve is the lender's pricing discipline made visible: for any estimated default probability, it gives the coupon required, against which the offered coupon can be judged.

Required coupon rises with default probability; the credit-spread component rises faster than the total.

Figure 3 translates the curve into discrete underwriting tiers, showing the required coupon for each of five risk tiers from low to high. The tiering is how the framework is used in practice: a transaction is assessed, assigned to a tier, and priced to the required coupon for that tier. A transaction offered at or above its tier's required coupon creates value; one offered below it, however high the absolute coupon, does not. This is the operational form of Proposition 3: the same offered coupon can clear the bar for a low tier and fall short for a high one, so the offered coupon is meaningless without the tier.

Figure 5. Risk-Based Required Coupon by Underwriting Tier

Each tier carries a required coupon; an offered coupon creates value only if it clears its tier's bar.

Default Probability by Sponsor and Asset

Implementation Considerations

Turning the framework into a live underwriting capability requires decisions about process, people, data and governance.

Building the Underwriting Process

The process should run every transaction through the same sequence: structured assessment of sponsor, asset and structure; independent estimation of default and recovery; computation of the required coupon; and comparison with the offered coupon. The sequence must be followed in order, with the risk assessment completed before the coupon is considered, so that the assessment is not contaminated by the price. A process that lets the underwriter see the coupon first will, consciously or not, reverse-engineer an assessment that justifies it, which defeats the purpose.

Structuring for Recovery

Because recovery is the highest-leverage variable, the structuring function deserves the most senior attention. The lender should seek first-ranking security, clear and enforceable title, completion and cost-overrun protections, cash-flow control and step-in rights, and should map the enforcement path before funding, not after default. Each structural feature that raises net recovery lowers both the expected loss and the capital charge, and so improves the economics more than an equivalent increase in the coupon. Structuring is examined in depth in the companion paper on security and enforcement.

Data and Calibration

A young market provides little historical loss data, which is a genuine constraint on calibration. The disciplined response is to calibrate conservatively, to record every transaction's assessment and outcome systematically so that a proprietary dataset accumulates, and to update the calibration as realised losses provide evidence. A lender that captures its own loss experience rigorously will, over a few years, hold a dataset that no newcomer can replicate, and that dataset is itself a source of edge. The absence of market data is a barrier to entry that rewards the lender who builds its own.

People and Judgement

The framework disciplines judgement; it does not replace it. The estimation of default probability, in particular, rests on experienced assessment of sponsors, assets and structures that no model can fully capture in a data-scarce market. The lender's underwriting talent, its ability to read a sponsor, value an asset and anticipate how a structure behaves in stress, is the ultimate determinant of loss experience. The framework ensures that talent is applied consistently and checked against the coupon; it does not manufacture the talent, which must be hired, developed and retained.

Governance and the Credit Committee

The credit committee is where the framework becomes governance. A committee presented with a transaction should see the underwriting score, the estimated default and recovery, the required coupon, the offered coupon and the margin between them, and should make its decision on that basis rather than on the headline yield. Presenting transactions in this decomposed form disciplines the committee as the scorecard disciplines the underwriter, and it creates an auditable record of why each transaction was approved at the price it was. Over time the record of decisions against outcomes is the institution's most valuable underwriting asset.

Pricing to Win Without Mispricing

A practical tension in any competitive lending market is between pricing to win the transaction and pricing to compensate for the risk. The framework resolves the tension by making the required coupon explicit: the lender may choose to price at or slightly above the required coupon to win a transaction whose risk it likes, but it should never price below the required coupon, because to do so is to lend at a loss in expectation. A lender that loses transactions because competitors price below the required coupon is not losing business it should regret; it is declining to fund losses that its competitors will eventually realise.

For Allocators: Assessing a Manager's Underwriting

For the allocator backing a Gulf credit manager rather than lending directly, the framework is a diligence tool. The allocator should ask the manager to demonstrate its underwriting process: how it estimates default and recovery, how it computes a required coupon, how it structures for recovery, how it capitalises unexpected loss, and how its realised losses compare with its underwriting scores. A manager that can answer these questions in decomposed, evidenced terms has the discipline the high-yield book requires; one that points to its coupons as evidence of its skill has confused price with risk, and is the manager most likely to deliver the loss the coupon was meant to compensate.

Common Underwriting Errors

Concluding Comments

This paper has set out a disciplined framework for underwriting and pricing high-yield Gulf real-asset credit, built around the two quantities that determine whether a coupon is sufficient: the probability of default and the loss given default. From these, together with the cost of capital and a target return, it derives the coupon a rational lender should require, and it argues that the lending decision should be driven by this required coupon, with the offered coupon serving as a check rather than a driver.

The central message is that underwriting, not yield, should drive lending. A high coupon is a question, whether it adequately prices the risk, and the answer differs across transactions that quote the same headline. A lender that estimates loss independently, structures to control recovery, capitalises unexpected loss and prices to a return on that capital will build a book whose realised losses are low and whose returns survive a cycle. A lender that lets the coupon drive the decision will, sooner or later, fund the transaction whose coupon was genuine compensation for a loss that duly arrives.

The framework is properly qualified. It is a structured, assumption-driven engine rather than a fitted statistical model, because the market lacks the data such a model would require; its figures are illustrative, not forecasts; and the estimation of default probability rests on judgement that no framework can fully systematise. These are limitations, but they are also the conditions that make disciplined underwriting valuable: in a data-rich, efficient market the edge would be competed away, while in a data-scarce, segmented one it accrues to the lender who underwrites best.

The study suggests several extensions. As realised loss data accumulate, the judgemental default and recovery estimates can be calibrated against outcomes, tightening the framework. The correlation structure of losses across the book, treated here through scenario analysis, could be modelled explicitly to size the unexpected-loss buffer more precisely. And the scorecard could be validated against realised losses to refine its weights. Each extension would sharpen the framework without changing its philosophy.

The framework also reframes what competitive advantage means in this market. In a mature, data-rich market, advantage comes from scale, cost of funding and access to flow; in a young, data-scarce one, it comes from underwriting judgement, the discipline to price for capital and not just loss, and the patience to decline transactions that competitors fund. These are capabilities a lender builds rather than buys, and they compound: each year of disciplined underwriting adds to the proprietary loss dataset, sharpens the calibration, and widens the gap between the disciplined lender and the yield-chaser. The advantage is durable precisely because it cannot be replicated with capital alone.