Chapter 4 · Credit Risk

16 of 100 exam questions (~16%) · counterparty vs issuer risk · EL = PD x EAD x LGD · ratings agencies · netting, collateral, CDS, securitisation · HHI · credit VaR
← Back to quiz
Why this chapter matters. 16 of 100 exam questions (~16%) on the v12 weighting, the second heaviest chapter (the Edition 11 workbook states 15; the bank uses v12 weights). It has the most calculations in the course: expected loss, recovery rate, netting, haircut collateral, HHI and the meaning of a credit VaR figure. Trap zones: counterparty risk (the other party fails) vs issuer risk (the bond issuer fails), RR = 100% - LGD, a CDS protection buyer swaps default risk for counterparty risk on the seller, wrong way risk, ratings measure credit quality only (not market or liquidity risk), and HHI ignores credit quality.

4.1 Identifying credit risk: counterparty, issuer, concentration

▼

The definition and the goal syllabus 4.1.1

Credit risk is the risk of loss caused by the failure of a counterparty or issuer to meet its obligations. The party owing the obligation is the obligor. The goal of credit risk management is to maximise a firm's risk-adjusted rate of return by keeping credit exposure within acceptable parameters. The original Basel Accord focused on credit risk because of its primary importance in many institutions' risk profiles. Techniques can be applied to a counterparty, an industry or a country, since default likelihood is strongly correlated with industry success and a home country's economy.

Counterparty risk vs issuer risk syllabus 4.1.1 - 4.1.2

FormMeaningWorkbook examples
Counterparty riskA counterparty (one of the parties to a transaction, buyer or seller) fails to fulfil its contractual obligationsA customer fails to repay a loan; a company doing business with the bank goes bankrupt before paying; a broker fails to deliver, or delivers late, a bond the bank bought; a broker agrees to buy shares, credits the client, and the client goes bankrupt before delivering; a bank's mortgage customers default
Issuer riskThe issuer of a bond or debt instrument defaults on coupons or principalA fixed income fund holding corporate bonds: if an issuer defaults, coupons stop and the fund's returns fall

The same bond purchase can carry both. If the broker fails to deliver a bond from a good issuer, that is counterparty risk. If the broker delivers exactly as required but the issuing company goes bankrupt and stops paying interest, that is issuer risk.

Memory hook: counterparty = the other side of MY trade fails; issuer = the company that issued the bond fails.
Quick check
A broker delivers a bond exactly as agreed, but the company that issued it then goes bankrupt. This is an example of:

Concentration risk syllabus 4.1.1

Concentration risk comes from an uneven distribution of exposures:

  • Single-name concentration: over-dependence on a small number of issuers or counterparties. If any default, revenues could drop significantly.
  • Sectoral concentration: uneven exposure within industry sectors and geographical regions. If the country or sector a bank is concentrated in suffers a downturn, its revenues suffer relative to better-diversified competitors.

4.2 Exposure sources, settlement, systemic risk + boundary issues

▼

Where credit exposure arises syllabus 4.1.2

For most banks loans are the largest source of credit risk, but exposure also arises in the banking book and trading book, on and off the balance sheet: commitments and guarantees, interbank transactions, financial instruments (futures, options, swaps, bonds) and the settlement of these and other transactions.

FirmTypical credit exposure
Fixed income fundIssuer credit risk on bonds and corporate debt held
BrokerCounterparty credit risk when trading with clients (e.g. a client fails to deliver shares it sold)
BankCounterparty credit risk when it issues mortgages to retail customers

Settlement risk vs pre-settlement risk syllabus 4.1.2

RiskWhenWorkbook example
Settlement riskAt the point buyer and seller exchange instrument and cash: one party delivers, the other fails toA forward exchanging euros for US dollars: on the maturity date the investment company pays euros but, because of time differences, the foreign bank's dollar payment is delayed. Ideally both transfers happen at the same time, which electronic settlement systems aim for, but the mechanism can fail
Pre-settlement riskBefore settlement: an institution defaults where the instrument has a positive economic value to the other partyAn interest rate swap (fixed vs floating payments on a specified principal amount): if rates move in favour of counterparty A, B owes a net obligation, and B's failure to perform creates pre-settlement risk for A
Trap: settlement risk is about the exchange moment (delivery vs payment); pre-settlement risk is about default earlier in the life of the contract while it is worth something to you.
Quick check
An interest rate swap counterparty defaults before the contract ends while the swap has positive value to you. This is:

Systemic risk in a credit context syllabus 4.1.2

Systemic risk is a possible breakdown of the entire financial system, not just one firm, arising from close interlinkages and feedback loops that can turn minor events into major crises. The workbook's contrast:

CaseWhat happenedSystemic?
Northern RockUK bank that borrowed cheaply on short-term interbank markets and turned it into long-term mortgages. The US housing credit crisis made banks withdraw from interbank lending. It failed in 2007-08, the first UK banking failure caused by deposit withdrawals for 140 yearsPart of a system-wide loss of trust
Barings (1995)Failed through a rogue trader's losses; bankrupt but no wider crisis because the causes were unique to that institutionNo contagion

Firms should form their own strategies rather than rely on governments to prevent systemic failure.

Credit risk boundary issues (Basel) syllabus 4.1.3

Basel identifies operational "boundary" risks to consider when banks develop their credit administration:

DriverBoundary risk
Internal processesEfficiency and effectiveness of credit administration: management policies and procedures for monitoring documentation, controls over back-office procedures, contractual requirements, legal covenants, collateral management
SystemsAccuracy and timeliness of credit risk information supplied to management information systems
PeopleAdequate segregation of duties

4.3 Measuring: exposure, premium, ratings

▼

Credit exposure syllabus 4.2.1

Credit exposure is the amount that can potentially be lost if a debtor defaults; it assesses severity for counterparties and portfolios. It has two parts:

PartMeaningDifficulty
Current exposureThe obligation currently outstandingNormally fairly straightforward to calculate
Potential future exposureEstimate of likely loss at some future pointHarder: credit facilities are often not drawn until the company is in trouble, and instruments' future values depend on circumstances not yet arisen (e.g. rate changes). Usually estimated with statistical techniques such as VaR

Credit risk premium and credit ratings syllabus 4.2.1

Credit risk premium = interest rate paid by a borrower - rate on a default-free security (e.g. a government bond)

It is the extra compensation lenders require for the risk of default. As an obligor's credit risk rises, lenders demand a higher premium. There is a strong relationship between premium and rating: higher rating = more creditworthy = lower premium = cheaper borrowing. Downgrades typically raise borrowing costs significantly.

A credit rating expresses creditworthiness and financial health. An independent agency assigns it using numeric factors (e.g. analysis of financial statements) and softer factors (management quality, competitiveness, expected growth of the industry). Agencies rate issuers and also individual issues, and issue ratings usually carry a short- and long-term outlook.

Memory hook: worse rating, higher premium, higher cost of borrowing.
Quick check
The credit risk premium is the difference between a firm's borrowing rate and the rate on:

Credit rating agencies: role and recognition syllabus 4.2.2

  • There are more than 75 ratings agencies worldwide, many specialising in regions or areas of credit finance.
  • The Basel Accord introduced a ratings-based method for credit risk in the standardised approach, and the Basel Committee defined guidelines for national regulators to recommend permissible agencies.
  • UK regulators nominated: Fitch, Moody's, Standard & Poor's (S&P) and DBRS. Germany's Bundesbank nominated those four plus the Japan Credit Rating Agency (JCRA).
  • Official recognition added to their influence, and investors relied heavily on the ratings. The workbook also cites Saudi Arabia's regulation of agencies' conduct.
  • Moody's, S&P and Fitch use similar but not identical letter scales (AAA is S&P's highest). By market convention, AAA to BBB = investment grade and BB to D = speculative grade. These are conventions and imply no recommendation of a security.
Trap: BBB is still investment grade; the speculative grade starts at BB.

Merits and limitations of ratings syllabus 4.2.3

Merits: a useful way for lenders to judge borrowers. Sovereign ratings reflect a country's economic conditions (foreign, public and private investment, capital market transparency, foreign currency reserves) and political conditions (stability, stability through transitions). Institutional investors and banks often use them as a prerequisite when deciding whether to look at companies, industries and securities in that country. Long-term ratings assess a company's ability to meet responsibilities across all its securities; short-term ratings focus on specific securities' ability to perform given current financial condition and industry conditions.

Limitations:

  • A rating measures credit quality only, not the risk of a fall in market value or liquidity prospects. Before the credit crunch many investors treated it as if it were, and relied on ratings rather than their own analysis.
  • A rating is an opinion (an expert one).
  • Slow downgrades: the workbook says Enron stayed investment grade until four days before bankruptcy, though agencies had known of problems for months. But ratings would lose stability if they tracked every market move.
  • Conflicts of interest: too familiar a relationship with company management, and a model where the receiver of the rating pays for it.
  • Errors on structured products: AAA ratings on structured debt later downgraded or defaulted; Fitch, Moody's and S&P publicly admitted they had not anticipated the credit crunch.
  • Over-correction in 2008-09: investment-grade ratings cut straight to speculative on an unprecedented scale though many bonds did not default; agencies accused of becoming too conservative, causing surges in banks' capital charges, because capital requirements depend on the rating.
Quick check
A credit rating of an instrument is designed to measure its:

4.4 Expected loss: PD, EAD, LGD, RR

▼

The expected loss formula syllabus 4.2.4

Expected loss (EL) = PD × EAD × LGD
TermUnitMeaning
PD probability of default%Likelihood a borrower (or group) fails to pay what is owed; estimated from historical experience and empirical evidence. Higher PD = higher interest rate charged
EAD exposure at defaultAmountThe amount the bank will be exposed to at the point of a potential default: may differ from today's exposure, e.g. facilities with undrawn commitments include an estimate of future lending before default
LGD loss given default%The percentage of the actual amount NOT mitigated by, e.g., guarantees or collateral
RR recovery rate%Amount likely to be recovered (recovery value) as a percentage; RR = 100% - LGD

Maturity: EAD depends on maturity (time to completion). The longer the time to maturity, the larger the chance credit quality falls, as the obligor has more opportunity, and perhaps need, to draw the remaining credit line.

Worked example (the workbook's Newbank case, plus an assumed PD)
Newbank lends ABC Co. $500 million. ABC defaults and the actual loss after guarantees and collateral is $300 million. Find LGD and RR. Then, assuming a PD of 2% (an assumption for this exercise, not from the workbook), find the expected loss.
  1. 1 LGD = actual loss / amount lent = 300 / 500 = 60%.
  2. 2 RR = 100% - LGD = 100% - 60% = 40% (about $200 million recovered).
  3. 3 EL = PD × EAD × LGD = 2% × $500m × 60% = $6 million.
Memory hook: EL = PD x EAD x LGD ("Probably Expensive And Large Damage" if you need one), and RR is the other side of LGD.
Quick check
If loss given default (LGD) is 60%, the recovery rate (RR) is:

4.5 Credit events, wrong way risk, NPAs, limits, model limitations

▼

Credit events syllabus 4.2.4

A credit event is a recognised industry trigger point telling firms a credit risk has materialised, triggering actions by the bank and other creditors. It has no precise definition but commonly includes bankruptcy, insolvency, receivership, or a failure to meet payment obligations when due. It can also be simply a credit-rating downgrade.

Wrong way risk syllabus 4.2.4

ISDA defines wrong way risk (or exposure) as the risk that occurs when exposure to a counterparty is adversely correlated with that counterparty's credit quality. It is missed if creditworthiness and exposure are modelled independently.

Example: accepting securities issued by the counterparty (or related entities) as collateral against that counterparty's default. The closer the counterparty gets to default, the lower the value of its equities, so the less cover they give, exactly when cover is needed.

Trap: collateral in the counterparty's OWN shares is the classic wrong way risk example.
Quick check
Which arrangement is the workbook's example of wrong way risk?

Non-performing assets syllabus 4.2.4

Non-performing assets are loans whose repayments are not being made on time. A short delay makes a loan past due; once payments are "really late" (usually 90 days) it is classified non-performing. A high level versus similar lenders, or a sudden increase, may signal problems, but judge it in context: banks lending to higher-risk customers tend to have more non-performing debt and compensate with higher interest rates, and a secured loan default is less concerning than an unsecured one because the security can be sold to recover the loss.

Credit limits syllabus 4.2.5

  • Credit limits are maximum limits for all aspects of credit exposure, set for individual borrowers, counterparties, and groups of connected counterparties. Better quality counterparty = higher limit, often based partly on internal risk rating.
  • Every institution sets them (bank, derivative dealer, fund manager, credit card company). They help keep credit granting adequately diversified.
  • Limits on trading book and off-balance-sheet activity are particularly effective for the overall credit risk profile and for linking to risk appetite.
  • Set limits also for industries or sectors, regions and products, and possibly a total limit for all corporate lending.
  • To be effective, limits should be binding and not driven by customer demand.
  • Trap: it is tempting to limit only lower-rated firms and lend without limit to higher-rated ones, but that increases concentration risk.

Limitations of credit risk measurement syllabus 4.2.6

LimitationIssue
Simplified potential exposurePotential exposure of a portfolio is generally greater than its current exposure; charges as percentages of notional amounts oversimplify, and notionals do not always measure true credit risk
No recognition of time periodDefault risk rises as exposure time increases, not always accounted for
No recognition of diversificationDiversification significantly reduces overall credit risk but calculations may ignore it
Probabilities are a best guessBased on past information, so PD can be miscalculated

4.6 Mitigation I: underwriting, guarantees, netting, collateral, diversification

▼

Underwriting standards and guarantees syllabus 4.3.1

ToolKey points
Underwriting standardsThe standards applied to borrowers to evaluate creditworthiness: review cash flow and financial statements; earnings, profit margin and outstanding debt; industry variables (competitive pressures, product cycles, growth potential); and controlling loan terms (limit loan size, repayment schedule, extra collateral for higher risk loans)
GuaranteesA strong-finance third party guarantees a debt (bond issuers, or smaller companies getting bank loans). The investor relies on the guarantor's creditworthiness and so takes counterparty risk on the guarantor. The borrower pays a cost but may gain lower-cost funding

Netting agreements syllabus 4.3.1

A netting agreement lets two parties who exchange multiple cash flows on a day agree bilaterally to net them to one payment per currency. It reduces settlement risk, transaction costs and communication costs. To actually reduce risk, agreements must be sound and legally enforceable.

The workbook's three-party example (A, B, C) shows the effect: if party C defaults with no netting, replacement costs would be £4m for A and £6m for B; with a netting agreement between each pair, they fall to £2m for A and £3m for B. Central counterparties also boost the scope for netting.

Exam approach: netting reduces the amount at risk on default and settlement risk. It only works if legally enforceable.
Quick check
A bilateral netting agreement mainly works by combining cash flows into:

Collateral syllabus 4.3.1

Collateral is an asset held by a lender on behalf of an obligor, on agreed conditions, as security for a loan: a physical asset (a house securing a mortgage), cash or securities. It works like insurance to reduce exposure, and on default the lender may retain it. It is used for FX forwards, securities lending and derivatives.

ArrangementMeaning
UnilateralOne party gives collateral to the other
BilateralTwo-sided obligations (swap, FX forward): both may post collateral for the value of their total obligation
NettedOnly the net obligor posts collateral, for just the net obligation

In a typical arrangement the collateral is periodically marked to market (revalued at current prices) and the amount adjusted.

Diversification syllabus 4.3.1

Spread the portfolio across borrowers in different, negatively correlated sectors with an inverse economic relationship. When one sector is in a downturn and its borrowers default, another may be booming, so earnings on some loans offset losses on others. The "highs" are lower and the "lows" are not as low, so the volatility of returns is lower. It helps avoid unacceptable concentrations.

4.7 Mitigation II: CDS, CDOs, securitisation, CCPs + haircut

▼

Credit derivatives and the credit default swap (CDS) syllabus 4.3.1

Credit derivatives derive value from an underlying loan or loans; the two most common are the CDS and the CDO. In a CDS the protection buyer pays a periodic or one-off fee (typically in basis points of the notional) to the protection seller, who pays out if a credit event occurs to the reference entity or reference asset. It resembles insurance.

Uses: buy or sell a form of insurance; improve diversification by cutting concentrations; customise exposure to another party without a direct relationship; transfer credit risk without harming the customer relationship. It also exposes the user to operational, counterparty, liquidity and legal risks.

Workbook example: Institution B buys bonds from Customer C (reference entity). B pays Institution A a fee for the life of the swap. If C defaults on a credit event, A pays B the default amount and the swap terminates. A assumes the credit risk, and B is now exposed to counterparty risk on A.

Trap: a CDS does not remove risk entirely; the buyer swaps risk on the borrower for counterparty risk on the seller.
Quick check
A bank buys CDS protection on a loan. After the trade, it is newly exposed to:

CDOs, loan sales and securitisation syllabus 4.3.1

ItemMeaning
CDOLike a CDS it shifts risk of an underlying debt instrument from lender to investor, but for a series of loans or bonds from many borrowers, achieving diversification because losses on one are absorbed by gains on non-defaulters
Loan saleLender sells loans to another institution for an immediate lump sum instead of relying on the income stream
SecuritisationThe originator owns the assets; an issuer (often a special purpose vehicle, SPV) acquires them and issues bonds to finance the purchase; loan income repays bondholders. The SPV holds the pool separately, so if the originator goes bankrupt the transferred assets are unaffected. Pools are often rated so a lower-rated originator can access cheaper debt capital markets, because a pool can attract a higher rating than the originator

The downside (2007-08): sub-prime mortgages were mixed with higher-rated ones in complex CDOs that earned higher ratings than the parts warranted, then re-cut and sold on. When rates rose in 2007 and sub-prime loans defaulted, the quality of securitisations could not be judged and demand collapsed. Institutions reliant on securitisation for funding were early casualties as inter-bank trust evaporated.

Central counterparties (CCPs) syllabus 4.3.1

A CCP (clearing house) is used by many exchanges to reduce credit risk. It acts as guarantor of all transactions, sitting as a simultaneous counterparty to each side, so members do not assume each other's credit risk; if one member defaults the clearing house guarantees performance to the other. CCPs also increase the scope for netting. To do this they need significant resources, from capital supplied by members, fees generated by the exchange, or other parties with no direct relationship to the market.

Collateral adequacy calculation: the haircut syllabus 4.3.2

When an investment manager lends stock (e.g. to firms wanting short positions), the borrower lodges collateral, often government bonds or other high-quality liquid assets. If the collateral's price falls while the stock is on loan, a default could leave the lender short, so an extra amount, a haircut, is added, depending on the collateral's volatility (for government bonds, say 3%).

Collateral required = value lent + (value lent × haircut %)
Worked example (from the workbook)
$100 million of IBM shares are lent and government bonds are offered as collateral with a 3% haircut. How much collateral must be lodged?
  1. 1 Value lent = $100m.
  2. 2 Haircut amount = $100m × 3% = $3m.
  3. 3 Collateral = $100m + $3m = $103m.
Quick check
£50m of stock is lent against collateral carrying a 4% haircut. The collateral that must be lodged is:

4.8 The credit risk function, Basel stages, scoring, stress tests, internal ratings

▼

The credit risk management function and escalation syllabus 4.3.3 - 4.3.4

The function ensures credit risk is properly managed, but does not own any risk itself. Duties include:

  • owning the credit policy and ensuring adherence
  • setting, monitoring and reviewing credit limits
  • assessing potential credit risk events
  • ensuring credit-granting decisions are made independently of the trading areas
  • measuring and monitoring daily credit exposure, providing information for capital adequacy assessment
  • credit analysis by counterparty, country, sector and instrument or product (which may include KYC checks)

The role varies by firm type: a bank trading for its own account can apply some limits across the board, while each fund in an investment manager can have a different risk appetite. Firms also need procedures and systems to monitor individual and grouped counterparties, with criteria for escalating potential credit issues to senior management. Problem areas need more frequent monitoring and possibly corrective action, classification and/or provisioning.

Basel: board, senior management and measurement syllabus 4.3.5

  • The board approves and reviews (at least annually) the credit risk strategy and significant credit risk policies, reflecting risk tolerance and expected profitability.
  • Senior management implements the strategy and develops policies and procedures to identify, measure, monitor and control credit risk across all activities.
  • Policies must be communicated, implemented via procedures, monitored and periodically revised. Banks should quantify risk for individual borrowers and analyse at product and portfolio level to spot sensitivities or concentrations.
  • Measurement should consider: (1) the nature of the credit and its conditions (maturity, reference rate); (2) the exposure profile to maturity relative to market movements; (3) collateral or guarantees, including any added counterparty risk; (4) potential default based on internal risk rating.
  • Techniques should suit the complexity and risk, use robust data and be subject to periodic validation: data quality assurance, model validation review, reviews of model assumptions.

The learning objective lists the stages as development, validation, approval, implementation, review and post-implementation monitoring; the workbook text covers them through the points above.

Credit scoring and factor inputs syllabus 4.3.6

Retail scoring: questionnaires and standard application forms, scored, using questions such as age, credit history, amount owed, occupation, years in current job, and home owner or renting. Firms are scored with three groups of factor inputs:

Input typeExamples
FinancialEarnings, cash flow, asset values, liquidity, leverage, financial size, debt capacity
Non-financialManagement quality, governance structure, industry characteristics, country risk, credit rating
ExtraordinaryCourt actions and other one-off factors that could affect ability to honour commitments
Trap: the credit rating and management quality are non-financial inputs; leverage and liquidity are financial inputs.
Quick check
Which of these is a NON-FINANCIAL input to a firm's credit score?

Stress testing and segmentation syllabus 4.3.6

ToolKey points
Stress testingIdentify possible events or changes in economic conditions that could hurt credit exposures and assess the bank's ability to withstand them. Areas the BIS says banks could "usefully examine": economic or industry downturns, interest rate and other market movements, market-risk events, liquidity conditions. Ranges from simple changes to one or more inputs to sophisticated models (typically large, internationally active banks). Review outputs periodically and act when results exceed tolerances, feed them into policies and limits, and use them for action plans such as hedging or reducing exposure
SegmentationUnder Basel, retail banking attracts less capital than commercial banking, and banks give regulators PD, LGD and EAD statistics for clearly differentiated portfolio segments. Segmentation is based on (i) credit scores (or equivalent) and (ii) time the transaction has been on the books

External vs internal ratings syllabus 4.3.6

ApproachPoints
External ratingsUseful but of limited value to a sophisticated credit function: often too historic, agencies slow to react to adverse events, not detailed or sensitive enough versus the firm's own analysis
Internal ratingsThe Basel Accord aims to incentivise banks to build their own systems. They differentiate the credit risk across exposures, and help judge portfolio characteristics, concentrations, problem credits and loan loss reserves. Larger banks also use them for internal capital allocation, pricing and profitability. They avoid split ratings (different agencies' ratings differing by several notches). Better systems have numerous gradations for satisfactory credits, and banks decide whether to rate the borrower, the transaction, or both. Ratings must be reviewed periodically and re-rated as conditions change, and set or confirmed by a credit review function independent of the one that originated the credit

Impairment and key statistics / KPIs syllabus 4.3.6

Evidence of impairment:

  • significant financial difficulty of the borrower (e.g. shown by liquidity or cash flow projections)
  • an actual breach of contract (e.g. late principal or interest payments)
  • a high probability of bankruptcy or reorganisation (e.g. a downgrade by a rating agency)
  • the lender granting a concession it would not otherwise consider, for economic or legal reasons related to the borrower's difficulties

Loans should be regularly reviewed for deterioration so impairment is identified in a timely way. A management dashboard should show key statistics and KPIs, such as the number of debtors past their due date, clients breaching covenants and credit downgrades.

4.9 Concentration (HHI) + credit VaR

▼

Controlling concentration risk syllabus 4.3.7

Single-name concentration is monitored centrally with lending limits per counterparty, with two challenges: (1) knowing the relationships between counterparties so that differently named entities in one corporate group are not missed; (2) loans may be only half the picture, as the bank could also hold equities in the firm, have lent stock to it, be managing assets for it and so on. Country and industry over-concentration also needs limits on lending across specific countries and sectors. Measuring concentration is the first step to managing it.

The Herfindahl-Hirschman Index (HHI) syllabus 4.3.7

HHI = sum of the squared relative (percentage) shares of each exposure in the portfolio

By construction the HHI lies between zero and one, with one meaning full concentration in a single counterparty. Advantages: simple to compute, moderate data needs. Disadvantage: it ignores credit quality (an exposure to an AAA borrower is treated like one to a B-rated borrower).

Worked example (from the workbook)
Two portfolios each total 50m. Portfolio A has two loans of 25m. Portfolio B has loans of 1m and 49m. Compute each HHI.
  1. 1 Portfolio A shares: 50% and 50%, i.e. 0.5 and 0.5. Squares: 0.25 + 0.25 = 0.50 (50%).
  2. 2 Portfolio B shares: 2% and 98%, i.e. 0.02 and 0.98. Squares: 0.0004 + 0.9604 = 0.9608 (96.08%).
  3. 3 Same total size, very different concentration: B's HHI is close to one, meaning almost fully concentrated in one loan.
Memory hook: the closer HHI is to 1, the more concentrated; more equal, more numerous positions push it towards 0.
Quick check
A portfolio has two equal loans, each 50% of the total. The HHI is:

Credit VaR and confidence levels syllabus 4.3.8

Value-at-Risk (VaR) is used to model credit risk in the trading book, at bond and portfolio level (developed further in the market risk chapter). The workbook explains the idea with a beach hut: measure a variable (tide height) over time, rank the readings, and read off the level exceeded only x% of the time (with 100 readings, the fifth highest is exceeded 5% of the time). Plotting the distribution and using standard deviation lets you read values at other probabilities.

For a bond, a commonly measured variable is its value in one year's time, based on the probability of moving from one credit rating to another (credit migration probabilities, published by rating agencies with the standard deviation). For a portfolio the same approach is used but correlations between bonds must be included to reflect diversification benefit.

The probabilities are confidence levels, the percentage that a value will not be exceeded. Example: a credit VaR of $5 million over one year at the 95% confidence level means there is only a 5% chance of losing more than $5 million in that time frame.

Trap: 95% confidence does NOT mean a 95% chance of losing $5m; it means a 95% chance of losing no more than $5m, i.e. a 5% chance of losing more.
Quick check
A portfolio's one-year credit VaR is $5m at the 95% confidence level. This means:

4.10 Ch 4 cheat sheet

▼

All the numbers + names

ItemAnswer
Credit riskLoss from failure of a counterparty or issuer to meet obligations (obligor = party owing)
Counterparty vs issuerParty to my transaction fails vs issuer of the security fails
ConcentrationSingle-name vs sectoral (industry, geographic)
Settlement vs pre-settlementAt exchange of cash and instrument vs default earlier while contract has positive value (e.g. swap)
Northern Rock / BaringsSystemic (deposit run) vs idiosyncratic (rogue trader 1995, no contagion)
Credit exposureCurrent exposure + potential future exposure (PFE estimated e.g. with VaR)
Credit risk premiumBorrowing rate minus default-free rate
GradesAAA-BBB investment; BB-D speculative
UK-recognised agenciesFitch, Moody's, S&P, DBRS (Germany adds JCRA)
Ratings limitsCredit quality only; opinion; slow (Enron); conflicts of interest; structured products
Expected lossPD × EAD × LGD
Recovery rate100% - LGD
Credit eventsBankruptcy, insolvency, receivership, missed payment; can be a downgrade
Wrong way riskExposure adversely correlated with counterparty credit quality (e.g. its own shares as collateral)
Non-performingUsually 90 days late (before that, past due)
Credit limitsBy borrower, counterparty, connected group, sector, region, product; binding, not driven by customers
Netting exampleDefault replacement costs £4m/£6m without vs £2m/£3m with
Collateral typesUnilateral · bilateral · netted; marked to market
CDSBuyer pays fee; seller pays on credit event; buyer gets counterparty risk on seller
SecuritisationOriginator → SPV/issuer → bonds; assets ring-fenced from originator's bankruptcy
CCPGuarantor of all trades; more netting
Haircut example$100m + 3% = $103m
Credit risk functionDoes not own risk; independent of trading; daily exposure monitoring
Board reviewCredit strategy and significant policies reviewed at least annually
Scoring inputsFinancial · non-financial · extraordinary
SegmentationBy credit score and time on books; retail attracts less capital than commercial
Internal ratingsIndependent credit review; avoid split ratings
HHISum of squared shares, 0 to 1; A = 0.50, B = 0.9608; ignores credit quality
Credit VaR$5m at 95% over one year = 5% chance of losing more than $5m