Pricing an NPL Portfolio: From Face Value to Expected Cash Flows

Pricing a non-performing loan portfolio is a cash-flow exercise. The face value of the claims is the starting perimeter, not the starting valuation, and the work consists of converting a list of legal claims into a schedule of expected receipts, net of what it costs to obtain them, positioned in time.

Why not a percentage of face value

Portfolios are still routinely discussed as a percentage of unpaid principal balance. As shorthand between people who already know the portfolio, that is harmless. As a pricing method it fails, because the percentage is an output of the analysis and not an input to it.

Two portfolios with identical face value can differ in every variable that matters: how much of the balance is capitalised interest that may not be recoverable as a matter of law, what proportion is secured and by what, how many borrowers are still trading, how far each file has progressed procedurally, and how concentrated the value is. A percentage carries none of that. It also imports whatever the last transaction cleared at, which may have been priced on a portfolio with nothing in common with this one.

Establish the perimeter

The first task is to establish what is actually being bought.

That means reconciling the tape to a defensible claim amount for each exposure: principal, contractual interest, default interest, capitalised amounts, fees and recoverable costs, tested against what would survive challenge. It means confirming which exposures are in the perimeter and which have been carved out. And it means identifying, in writing, what the data does not support — the missing valuations, the unregistered security, the exposures where procedural status is unknown.

That last document is not an administrative artefact. It determines how much of the portfolio can be underwritten exposure by exposure and how much has to be treated statistically, which in turn determines how conservative the pricing has to be.

Segment, then find the concentration

Segmentation groups exposures that will behave alike: by borrower type, exposure size, security status, collateral type, procedural stage and payment behaviour since default.

The purpose is not tidiness. It is to locate the concentration. In most portfolios a small proportion of exposures accounts for the majority of the recoverable value, and that subset has to be underwritten individually. The remainder can be modelled on segment assumptions, provided the segments are homogeneous enough that an average within them means something.

The practical test is straightforward: if the top exposures by expected recovery are underwritten file by file, the portfolio is being priced. If they are being modelled on a segment curve, it is being estimated.

Assign recovery paths

Each segment and each material exposure is assigned the recovery routes actually open to it. In practice these are: voluntary payment, negotiated settlement, restructuring where the borrower is still operating, enforcement against collateral, enforcement against guarantors, and sale of the claim.

Each route carries its own expected amount and its own timetable, and the weighting between them should reflect the specific situation, not a portfolio-wide split. A borrower still trading with a viable business and a cooperative posture is a restructuring candidate; the same balance owed by a dissolved company with a mortgaged property is a collateral realisation with a court timetable attached. Averaging those two into a blended recovery assumption destroys the information that made the analysis worth doing.

Deduct the costs

Costs fall into two groups, and they behave differently.

Costs that scale with recovery: success fees, brokerage on disposals, taxes on transfer. These can be modelled as a percentage of the relevant cash flow.

Costs that scale with time and file count: servicing fees, legal fees, court and enforcement costs, valuation, and the holding costs of any repossessed asset. These accrue against files that remain open, and they must be modelled period by period rather than as a single deduction, because a slower timetable increases them.

Modelling all costs as one percentage of gross recoveries is the most common shortcut in NPL pricing and it biases the result in a predictable direction: it flatters slow portfolios, which are exactly the ones that need the discipline.

Discount at the required return

The net cash flows are then discounted at the return the investor requires for this risk, in the currency the investor funds in.

Three points are worth being explicit about. The discount rate is a decision, not a market observation: it belongs to the investor and should be stated, not derived. Where the investor funds in a currency other than the dirham, the cash flows and the rate have to be treated consistently, and the cost of hedging — or the decision not to hedge — is a real line in the model. And the output is not a valuation. It is a maximum acquisition price: the highest price at which this cash-flow schedule still clears the threshold.

Presenting that price at a single required-return level hides the trade-off. A price-to-return grid — the bid at several thresholds — is more useful to a committee, because it shows how much return is being given up for each increment of price.

Test the downside, then bid

The final step is to find out what the price depends on.

A small number of assumptions carry most of the value in almost every portfolio: the realisation on the largest collateral positions, the enforcement timetable, and the outcome on the top exposures by expected recovery. Those are tested individually, then together, in a coherent adverse scenario rather than a uniform haircut. A downside case in which enforcement takes three years longer, one significant security interest fails, and disposal proceeds come in below appraisal is a scenario. A flat twenty per cent reduction applied to the base case is not; it tells you nothing about which risk you are actually taking.

The bid then follows from two numbers rather than one: the price at which the base case meets the required return, and the price at which the downside case still avoids a loss. Where those two numbers are far apart, the portfolio is a bet on a specific assumption, and the memorandum should say which one.

Further reading

• How to Value a Non-Performing Loan Portfolio in Morocco
• Why Recovery Timing Matters as Much as Recovery Rate
• Collateral Value Is Not Recovery Value