Potential future exposure (PFE) is an estimate of the largest credit exposure a firm could have to a counterparty at a future date, or over the remaining life of a portfolio of trades, measured at a high statistical confidence level such as 95 or 99 per cent. It applies mainly to derivatives and securities financing trades, whose value to each side changes with market prices until they mature. Banks use it to set credit limits, regulators use versions of it to calculate capital, and clearing houses apply the same idea when sizing margin.
PFE is one of two components of counterparty credit exposure. The other is current exposure, the loss that would arise if the counterparty defaulted today, equal to the positive replacement value of the trades. PFE adds an allowance for how far that value could rise before the trades can be closed out or replaced. The PFMI use the same split and require a central counterparty (CCP) to cover both current and potential future exposure to its participants with a high degree of confidence.
How potential future exposure is measured
Banks with advanced models usually estimate PFE by Monte Carlo simulation. Thousands of possible paths are generated for interest rates, exchange rates, credit spreads and other risk factors. At each date on a future time grid, every trade is revalued on every path, netting and collateral agreements are applied, and negative values are set to zero, since a counterparty that owes nothing causes no credit loss. The PFE at that date is a high percentile of the resulting distribution. Plotted against time it forms an exposure profile, whose highest point is called peak PFE.
The same simulation yields expected exposure (EE), the average of the positive exposures at each date, which feeds the pricing of credit valuation adjustment (CVA). PFE describes a bad but plausible case rather than an average one, so it is the usual measure for limits.
A simple model shows the idea. If a position starts with a value of zero and its changes in value are normally distributed with annual standard deviation , its PFE at horizon is approximately:
Here is the standard normal value for the chosen confidence level , which is 1.645 at 95 per cent, and is the horizon in years. For a forward contract whose value has an annual standard deviation of 2 million, the 95 per cent PFE after three months is million. After one year it is million.
What shapes the exposure profile
- Product type. An interest rate swap typically shows a hump: uncertainty about rates grows over time while the number of remaining payments shrinks. A currency forward or cross-currency swap with a final exchange of principal shows exposure that keeps rising to maturity.
- Netting. An enforceable close-out netting agreement lets gains and losses on different trades offset, often reducing PFE sharply compared with treating trades one by one.
- Collateral. Under a margin agreement, exposure can grow only over the margin period of risk, the time between the last collateral exchange and the close-out of trades after a default.
- Wrong-way risk. Where exposure tends to rise as the counterparty weakens, as with specific wrong-way risk, standard models understate the danger.
Regulatory add-ons and SA-CCR
Bank capital rules from the Basel Committee on Banking Supervision use a simpler, formula-based PFE. The older current exposure method added a fixed percentage of each trade's notional amount to its replacement cost. Its successor, the standardised approach for counterparty credit risk (SA-CCR), published in 2014, calculates exposure at default (EAD) for each netting set as:
- is a supervisory factor fixed at 1.4.
- is the replacement cost: the current market value of the netting set less net collateral held, floored at zero.
- is an aggregate add-on built from supervisory factors for each asset class, multiplied by a factor of at most one that gives credit for excess collateral or a negative market value.
For an unmargined netting set worth 2 million to the bank, with no collateral and an aggregate add-on of 3 million, the multiplier is one and the exposure is million. Banks with approved internal models instead multiply a simulated measure of expected exposure by a similar alpha factor.
Uses in credit limits and clearing
Credit departments commonly express a counterparty credit limit in PFE terms. A desk may trade with a counterparty only while the portfolio's peak PFE stays below the limit, and new trades are checked for their incremental effect before execution.
A CCP applies the same logic to each clearing member. Its initial margin is effectively a PFE estimate over the margin period of risk: under Principle 6 of the PFMI, initial margin should cover potential future exposure with a single-tailed confidence level of at least 99 per cent for each product.
Limitations
PFE depends heavily on models. Results vary with the assumed distribution of risk factors, the confidence level, the calibration period and the treatment of collateral, so two banks can produce very different figures for the same portfolio, an instance of model risk. Calibrations based on calm periods understate exposure just before stress, while calibrations that react quickly to volatility can add to procyclicality. Finally, PFE measures exposure only; it must be combined with the counterparty's probability of default and expected recovery to estimate credit loss.