HWESER

C r it ic a l C o n c e pt s for the 2019 FRM® E x a m

MARKET RISK MEASUREMENT Step 2 : Determine the ranks o f X. and Y for each Bond Valuation Using Binomial Tree
AND MANAGEMENT
time period i. Using backward induction, the value o f a bond at
Step 3 : Calculate the difference o f the variable a given node in a binomial tree is the average o f the
rankings and square the difference. present values of the two possible values from the
Value at Risk (VaR) n
VaR for a given confidence level occurs at the
cutoff point that separates the tail losses from the
6E di
i= i
next period. The appropriate discount rate is the
forward rate associated with the node under analysis.
There are three basic steps to valuing an option on
remaining distribution. Ps = l —
n(n2 - l ) a fixed-income instrument using a binomial tree:
H istorical simulation approach: order return
Where n is the number o f observations for each Step 1: Price the bond value at each node using
observations and find the observation that
variable and d.I is the difference between the the projected interest rates.
corresponds to the VaR loss level.
ranking for period i. Step 2 : Calculate the intrinsic value o f the
Parametric estimation approach: assumes a
derivative at each node at maturity.
distribution for the underlying observations. Kendall’s t
Step 3 : Calculate the expected discounted value o f
• Normal distribution assumption: nc ~ n j
T the derivative at each node using the risk-
VaR = (—p r + or x za ) n(n —1) / 2
neutral probabilities and work backward
Where the number o f concordant pairs is through the tree.
• Lognormal distribution assumption:
represented as n; (pair rankings in agreement),
VaR = ( l - e^ - * r xza) Interest Rate Expectations
and the number o f discordant pairs is represented
Expectations play an important role in
Expected Shortfall as n (pair rankings not in agreement).
determining the shape o f the yield curve and
Provides an estimate o f tail loss by averaging the Mean Reversion can be illustrated by examining yield curves that
VaRs for increasing confidence levels in the tail. • Implies that over time variables or returns regress are flat, upward-sloping, and downward-sloping.
back to the mean or average return.
Weighted Historical Simulation If expected 1-year spot rates for the next three
• Mean reversion rate, a, is expressed as:
years are r(, r2, and r3, then the 2-year and 3-year
Approaches S t — S t_ ! = a(p —S t_ i )
• Age-weighted: adjusts the most recent (distant) spot rates are computed as:
• The [3 coefficient of a regression is equal to the
observations to be more (less) heavily weighted. negative of the mean reversion rate. r(2) = yj(1 + q)(l + r2) —1
• Volatility-weighted: replaces historic returns with
volatility-adjusted returns; actual procedure of Autocorrelation
f(3 )= 3 /(l + r1)(l + r2)(l + r3) - l
estimating VaR is unchanged. • Measures the degree that a variable’s current value
is correlated to past values.
• Correlation-weighted: updates the variance-
• Has the exact opposite properties of mean
Convexity Effect
covariance matrix between assets in the portfolio. All else held equal, the value o f convexity
• Filtered historical simulation: relies on reversion.
• The sum of the mean reversion rate and the one- increases with maturity and volatility.
bootstrapping of standardized returns based on
volatility forecasts; able to capture conditional period autocorrelation rate will always equal one. Term Structure Models
volatility, volatility clustering, and/or data Correlation Swap M odel 1: assumes no drift and that interest rates
asymmetry. • Used to trade a fixed correlation between two or are normally distributed:
more assets with a realized correlation. dr = odw
Backtesting VaR
• Realized correlation for a portfolio of n assets:
• Compares the number of instances when losses exceed M odel 2 : adds a positive drift term to Model 1
the VaR level (exceptions) with the number predicted Prealized that can be interpreted as a positive risk premium
by the model at the chosen level of confidence. associated with longer time horizons:
• Failure rate: number of exceptions/number of dr = Xdt + odw
observations. • Payoff for correlation swap buyer:
• The Basel Committee requires backtesting at the notional amount x (prealized —pfixed) where:
99% confidence level over one year; establishes zones X = interest rate drift
for the number of exceptions with corresponding Gaussian Copula
• Indirectly defines a correlation relationship Ho-Lee Model: generalizes drift to incorporate
penalties (increases in the capital multiplier).
between two variables. time-dependency:
Mapping • Maps the marginal distribution of each variable dr = X(t)dt + odw
M apping involves finding common risk factors to a standard normal distribution (done on
Vasicek Model: assumes a mean-reverting process
among positions in a given portfolio. It may be percentile-to-percentile basis).
for short-term interest rates:
difficult and time consuming to manage the risk • The new joint distribution is a multivariate
dr = k(0 - r)dt + odw
o f each individual position. One can evaluate the standard normal distribution.
• A Gaussian default time copula can be used for where:
value o f portfolio positions by mapping them
measuring the joint probability of default between k = a parameter that measures the speed o f
onto common risk factors.
two assets. reversion adjustment
Pearson Correlation Coefficient 0 = long-run value o f the short-term rate
Commonly used to measure the linear Regression-Based Hedge assuming risk neutrality
relationship between two variables: DV01n ) r = current interest rate level
Fr = Fn x X3
DV01 r M odel 3 : assigns a specific parameterization o f
CQVXY
PXY time-dependent volatility:
aXaY
where: dr = X(t)dt + oe~atdw
F R = face amount o f hedging instrument where:
Spearman’s Rank Correlation F N = face amount o f initial position a = volatility at t = 0, which decreases
Step 1: Order the set pairs o f variables X and Y
exponentially to 0 for a > 0
with respect to set X.
Cox-Ingersoll-Ross (CIR) model: mean-reverting Exposure a t default (EAD): amount o f money Credit Risk Portfolio Models
model with constant volatility, a, and basis-point the lender can lose in the event o f a borrower’s These models attempt to estimate a portfolio’s credit
volatility, a V r , that increases at a decreasing rate: default. value at risk. Credit VaR differs from market VaR
dr = k(0 —r)dt + aVrdw Rating Systems in that it measures losses that are due specifically to
A good rating system has three key features: default risk and credit deterioration risk.
M odel 4 (lognormal model): yield volatility,
• Objectivity and homogeneity, produce judgments CreditRisk+: determines default probability
cr, is constant, but basis-point volatility, or,
tied solely to credit risk and ratings that are correlations and default probabilities by using a
increases with the level o f the short-term rate. comparable across market segments, portfolios, set o f common risk factors for each obligor.
There are two lognormal models o f importance: and customer types. CreditMetrics: uses historical data to estimate
(1) lognormal with deterministic drift and • Specificity, accurately capture the distance to the probability o f a bond being upgraded or
(2) lognormal with mean reversion. default while ignoring other non-default-related downgraded using historical transition matrices.
Put-Call Parity financial elements. K M V Porfolio M anager: default probability is
c - p = S - Xe-rT • Measurability and verifiability, provide accurate
a function o f firm asset growth and the level o f
expectations tied to default probabilities
where: debt. The higher the growth and lower the debt
backtested on a continuous basis.
c = price o f a call level, the lower the default probability.
Rating Assignment Methodologies CreditPorfolioView: multifactor model for
p = price o f a put
Experts-based approaches rely on experienced simulating joint conditional distributions o f
S = price o f the underlying security
individuals who can provide valuable inputs into credit migration and default probabilities that
r = risk-free rate
the models. incorporates macroeconomic factors.
T = time left to expiration expressed in years
Statistical-based models use both quantitative and
Volatility Smiles Credit Derivatives
qualitative data to describe the real world in a
Currency options: implied volatility is lower for A credit derivative is a contract with payoffs
controlled environment.
at-the-money options than it is for away-from- contingent on a specified credit event. Credit
Numerical-based models are designed to derive
the-money options. If the implied volatilities for events include:
optimal solutions using trained algorithms.
• Failure to make required payments.
actual currency options are greater for away-from- Linear discrim inant analysis (LDA) is used to • Restructuring that harms the creditor.
the-money than at-the-money options, currency develop scoring models (e.g., Altman’s z-score) to • Invocation of cross-default clause.
traders must think there is a greater chance o f provide accept/reject decisions. • Bankruptcy.
extreme price movements than predicted by a L O G IT models are used to predict default based Credit default swap (C D S): like insurance;
lognormal distribution. on understanding the relationships between party selling the protection receives a fee, pays
Equity options: higher implied volatility for low dependent and independent variables. based on swap’s notional am ount in the case o f
strike price options. The volatility smirk (half- Cluster analysis aggregates and segments default.
smile) exhibited by equity options translates into borrowers based on the profiles o f their variables. First-to-defaultput: C D S variation where a party
a left-skewed implied distribution o f equity price Principal component analysis takes original data pays an insurance premium in exchange for
changes. This indicates that traders believe the and transforms it into a new derived data set, being made whole for the first default from a
probability o f large down movements in price which is used to determine the primary drivers o f basket o f assets. More cost effective option than
is greater than large up movements in price, as a firm’s profile and potential default. C D S if assets have uncorrelated default risks.
compared with a lognormal distribution. Cash flow analysis is useful for assigning ratings to Total return swap: total return on an asset (bond)
companies that don’t have meaningful historical is exchanged for a fixed (or variable) payment;
data for predicting potential default. total return receiver gets any appreciation
The Merton Model (capital gains and cash flows), pays any
• A value-based model where the value of the firm’s depreciation; payments take place whether or
outstanding debt (D) plus equity (E) is equal to not a credit event occurs. Buyer exchanges credit
Credit Risk the value of the firm ( V). risk o f issuer defaulting for the combined risk o f
• Credit risk is either the risk of economic loss from • The value of the debt can serve as an indicator of the issuer and the derivative counterparty.
default, or changes in credit events or credit ratings. firm default risk. Vulnerable option: option with default risk; holder
• Types of credit risky securities include: corporate • Since E and D are contingent claims, option pricing receives promised payment only if seller o f the
and sovereign debt, credit derivatives, and structured can be used to determine their values as follows:
option is able to make the payment.
credit products. Their interest rates include a credit payment to shareholders: max(VM- D , 0)
Asset-backed credit-linked note: embeds a default
spread above credit risk-free securities.
payment to debtholders: swap into a debt issuance. It is a debt instrument
Classifications of Credit Risk D m - m a x (D M" V m’ ° ) with its coupon and principal risk tied to an
Risks relating to default include default risk, underlying debt instrument (e.g., bond or loan).
• Equity is similar to a long call option on the value
recovery risk, and exposure risk. Risks relating to
of a firm’s assets where face value of debt is the Spread Conventions
valuation include migration risk, spread risk, and
strike price of the option. Yield spread: YTM risky bond —YTM benchmark
liquidity risk. • Debt is similar to a risk-free bond and short put government bond
For revolving credit facilities, exposure risk option on the value of a firm’s assets where face i-spread: YTM risky bond - linearly interpolated
depends on borrower behavior and external value of debt is the strike price of the option. YTM on benchmark government bond
events. In this case, exposure at default (EAD) =
Credit Spread z-spread: basis points added to each spot rate on a
drawn amount + (limit —drawn amount) x loan
Difference between the yield on a risky bond benchmark curve
equivalency factor.
(e.g., corporate bond) and the yield on a risk- C D S spread: market premium o f C D S o f issuer
Expected Loss (EL) free bond (e.g., T-bond) given that the two bond
Expected value o f a credit loss: instruments have the same maturity.
EL = PD x L G D x EAD
Hazard Rates
1 The hazard rate (default intensity) is represented
CS = - F
Probability o f default (PD): likelihood that a (T-t) by the (constant) parameter X and the probability
borrower will default within a specified time where: o f default over the next, small time interval, dt,
horizon. D = current value o f debt is Xdt.
Loss given default (LG D ): amount o f creditor loss F = face value o f debt
in the event o f a default. In percent terms, it is
equal to 1 minus the recovery rate (i.e., 1 —RR).
Cumulative PD Recovery: measured by the recovery rate, which Credit Value Adjustment (CVA)
If the time o f the default event is denoted t*, is the portion o f the outstanding claim actually Expected value or price o f counterparty credit
the cumulative default time distribution F (t) recovered after default. risk. A positive value represents a cost to the
represents the probability o f default over (0, t): Wrong-way exposures: exposures that are negatively counterparty that bears a greater propensity
P(t* < t ) = F(t) = 1 - e-Xt correlated with the counterparty’s credit quality. to default. The CVA should account for the
They increase expected credit losses. counterparty’s default probability, the transaction
The survival distribution is: M ark-to-market (M tM ): accrual accounting in question, netting, collateral, and hedging.
P(t* > t ) = l - F(t) = e-Xt measure that is equal to the sum o f the M tM
values o f all contracts with a given counterparty. CVA = L G D x E x E E ( t i ) x P D ( t i _ 1, q )
Probability of Default i= l
The probability o f default (PD) o f a debt security
Credit Exposure Metrics
Expected M tM : forward or expected value o f a where:
can be calculated by using the following equation:
transaction at a given point in the future. EE = discounted expected exposure
Expected exposure (EE): amount that is expected Debt Value Adjustment (DVA)
LGD
to be lost (positive M tM only) if the counterparty Financial institutions should incorporate the
defaults. value o f their option to default to a counterparty
Collateralized Debt Obligations
Potentialfuture exposure (PFE): worst exposure through the bilateral CVA (BCVA), also known
• General term for an asset-backed security that
issues securities that pay principal and interest from that could occur at a given time in the future at a as the DVA.
a collateral pool of debt instruments. given confidence level. Unlike the CVA formula, the BCVA incorporates
• In order to create a CD O , the issuer packages a Expected positive exposure (EPE): average EE negative expected exposure and the probability o f
series of debt instruments and splits the package through time. the counterparty’s survival.
into several classes of securities called tranches. Effective EE: equal to non-decreasing EE.
• The largest part of a CD O is typically the senior Effective EPE: average o f effective EE.
Risk-Neutral Default Probability
tranche, which usually carries an AA or AAA credit Represents estimates o f default probability based
M aximum PFE: highest PFE value over a stated
rating, regardless of the quality of the underlying on observed market prices o f securities (e.g.,
time frame.
assets in the pool. bonds, credit default swaps),
Synthetic CD O : originator retains reference assets
Credit Mitigation Techniques risk-neutral default probability = real-world
Netting: a legally binding agreement that enables default probability + liquidity premium + default
on balance sheet but transfers credit risk to an
counterparties with multiple derivative contracts risk premium
SPV, which then creates the tradable synthetic
to net their obligations (e.g., Party A owes Party B
C D O . This product bets on the default o f a pool
o f assets, not on the assets themselves.
$50 million; Party B owes Party A $40 million, so Wrong-Way Risk vs. Right-Way Risk
Party A pays a net $10 million to Party B). Wrong-way risk: increases counterparty risk
Securitization Collateralization: if the value o f derivative [increases credit value adjustment (CVA) and
Transforms the illiquid assets o f a financial contracts is above a stated threshold, collateral decreases debt value adjustment (DVA)].
institution into a package o f asset-backed must equal the difference between the value o f Right-way risk: decreases counterparty risk
securities (ABSs) or mortgage-backed securities the contracts and the threshold level. Collateral (decreases CVA and increases DVA).
(M BSs). A third party uses credit enhancements, agreements have two major benefits: (1) expand
liquidity enhancements, and structuring to issue
OPERATIONAL AND INTEGRATED
the list o f potential counterparties because credit
securities backed by the pooled cash flows (of the rating is a less important concern and (2) reduce

RISK MANAGEMENT
same underlying assets). economic capital requirements.
• Credit enhancements include overcollateralization,
subordinating note classes, margin step-up, and Modeling Collateral
excess spread. Certain parameters impact the effectiveness o f Risk-Adjusted Return on Capital
• The first-loss piece (equity piece) absorbs initial collateral in lessening credit exposure. These The RA RO C measure is essential to successful
losses and is often held by the originator. parameters are as follows: integrated risk management. Its main function
Remargin period, the time between the call for is to relate the return on capital to the riskiness
ABS/MBS Performance Tools
collateral and its receipt. o f firm investments. The RARO C is risk-
Auto loans: loss curves, absolute prepayment speed.
Threshold: an exposure level below which adjusted return divided by risk-adjusted capital
Credit card debt: delinquency ratio, default ratio,
collateral is not called. It represents an amount o f (i.e., economic capital).
monthly payment rate.
uncollateralized exposure. RA RO C =
Mortgages: debt service coverage ratio, weighted
average coupon, weighted average maturity, M inimum transfer amount, the minimum quantity revenues —costs —EL —taxes +
weighted average life, single monthly mortality, or block in which collateral may be transferred. return on economic capital ± transfers
constant prepayment rate, Public Securities Quantities below this amount represent
economic capital
Association. uncollateralized exposure.
In itial margin: an amount posted independently An adjusted RA RO C (ARAROC) measure was
Subprime Mortgage Market o f any subsequent collateralization. This is also
• Subprime borrowers have a history of either default developed to better align the risk o f the business
referred to as the independent amount. with the risk o f the firm’s equity.
or strong indicators of possible future default.
Rounding, the process by which a collateral call Adjusted RARO C = RARO C - (3E(RM- R f)
• Indicators of future default: past delinquencies,
judgments, foreclosures, repossessions, charge-offs, amount will be adjusted (rounded) to a certain
increment.
Model Risk
and bankruptcy filings; low FICO scores; high debt Risk associated with using financial models to
service ratio of 50% or more. Central Counterparties (CCPs) simulate complex relationships. Sources o f model
• The vast majority of subprime loans are adjustable CCPs step in the middle o f a bilateral risk include model errors, errors in assumptions,
rate mortgages. counterparty relationship between member firms and errors in implementation. Com m on model
Counterparty Risk and offset risk with loss mutualization, collateral errors include underestimating risk factors,
The risk that a counterparty is unable or posting, and multilateral trade netting. They misapplying a model, and assuming constant
unwilling to live up to its contractual obligations. utilize a loss waterfall to mutualize losses among volatility, normal distributions, perfect markets,
Credit exposure: loss that is “conditional” on the all member firms. Losses are first absorbed by the and adequate liquidity.
counterparty defaulting. defaulted members’ initial margin and default
funds. If losses are greater, C C P equity and
Liquidity Risk
The lack o f a market for a security to prevent it
surviving members’ default funds are used.
from being bought or sold quickly enough to
 prevent or minimize a loss. It could result from Leverage Ratio • From the perspective of the borrower, repos offer
asset allocation, funding strategies, collateral A firm’s leverage ratio is equal to its assets divided relatively cheap sources of short-term funds.
policies, or mismanagement o f risks. by equity: • From the perspective of the lender, reverse repos are
Transactions liquidity risk: risk that the act o f used for either investing or financing purposes.
l = A = (E+D) = 1 + D
buying or selling an asset will result in an adverse E E E Capital Plan Rule
price move. • Mandates that bank holding companies develop a
Funding liquidity risk: results when a borrower’s Leverage Effect capital plan and evaluate capital adequacy.
credit position is either deteriorating or Return on equity (ROE) is higher as leverage • Capital adequacy process includes: risk
is perceived by market participants to be increases, as long as the firm’s return on assets management foundation, resource and loss
deteriorating. (ROA) exceeds the cost o f borrowing funds. The estimation methods, impact on capital adequacy,
leverage effect can be expressed as: capital planning and internal controls policies, and
Liquidity-Adjusted VaR (LVaR) RO E = (leverage ratio x ROA) — governance oversight.
The constant spread approach calculates liquidity-
adjusted VaR (LVaR) assuming the bid-ask spread
[(leverage ratio - 1) x cost o f debt] Operational Risk Governance
is constant. Transaction Cost The Basel Committee recognizes three common
The 99% confidence interval on transaction cost is: lines o f defense used to control operational risks:
LVaR = ( V x z a x o ) + (0.5 x V x spread)
LVaR = VaR + liquidity cost + /- P x V2(s + 2.33 ct s) (1) business line management, (2) independent
operational risk management function, and
where: where:
(3) independent reviews o f operational risks and
V = asset value P = estimate o f the next day asset midprice risk management.
za = VaR confidence parameter s = bid-ask spread
a = standard deviation o f returns p2 (s + 2.33(sj = 99% spread risk factor
Risk Appetite Framework (RAF)
• Sets in place a clear, future-oriented perspective of
Operational Risk Data Elements the firm’s target risk profile in a number of different
. (ask price —bid price) The four data elements that a bank must use in scenarios and maps out a strategy for achieving that
spread = --------------- :------------
(ask price + bid price) / 2 various combinations to calculate the operational risk profile.
risk capital charge are (1) internal loss data, (2) • Should start with a risk appetite statement that
external loss data, (3) scenario analysis, and (4) is essentially a mission statement from a risk
LVaR can also be calculated given the
business environment and internal control factors perspective.
distributional characteristics o f the spread. This
• Benefits include assisting firms in preparing for the
is known as the exogenous spread approach. If you (BEICFs).
unexpected and greatly improving a firm’s strategic
are given the mean and standard deviation o f the Basel II Operational Risk Event Types planning and tactical decision-making.
spread, apply the following formula: • Internal Fraud.
LVaR = VaR + 0.5 x (p$ + x )x V • External Fraud.
Basel II: Three Pillars
• Employment Practices and Workplace Safety. P illar 1: M inimum capital requirements. Banks
where: • Clients, Products, and Business Practices. should maintain a minimum level o f capital to
ps = spread mean • Damage to Physical Assets. cover credit, market, and operational risks.
(7S = spread standard deviation • Business Disruption and System Failures. P illar 2 : Supervisory review process. Banks should
z[x = spread confidence parameter • Execution, Delivery, and Process Management. assess the adequacy o f capital relative to risk, and
External Loss Data supervisors should review and take corrective
Liquidity at Risk (LaR) IB M Algo FIRST: subscription database; includes action if problems occur.
• Maximum likely cash outflow over the horizon P illar 3 : M arket discipline. Risks should be
descriptions and analyses o f operational risk
period at a specified confidence level. adequately disclosed in order to allow market
events derived from legal and regulatory sources
• Also known as cash flow at risk (CFaR). participants to assess a bank’s risk profile and the
and news articles.
• A positive (negative) value for LaR means the adequacy o f its capital.
worst outcome will be associated with an outflow Operational Riskdata eXchange Association
(inflow) of cash. (ORX): consortium-based risk event service; Basel II: Forms of Capital
• LaR is similar to VaR, but instead of a change in gathers anonymous operational risk events from Tier 1: shareholder’s equity, retained earnings;
value, it deals with cash flows. members. nonredeemable, noncumulative preferred stock.
Loss Distribution Approach (LDA) Tier 2 : undisclosed reserves, revaluation reserves,
Enterprise Risk Management (ERM)
The LDA is used to meet the Basel II operational general provisions/general loan-loss reserves,
In developing an ERM system, management
risk standards for regulatory capital. It relies on hybrid debt capital instruments, and subordinated
should follow the following framework:
• Determine the firm’s acceptable level of risk. internal losses as the basis o f its design. term debt.
• Based on the firm’s target debt rating, estimate the M odelingfrequency, first step is to determine the Credit Risk Capital Requirements
capital (i.e., buffer) required to support the current likely frequency o f events on an annual basis. The standardized approach incorporates
level of risk in the firm’s operations. Most common approach is Poisson distribution. risk weights based on external credit rating
• Determine the ideal mix of capital and risk that M odeling severity, next step is to determine the assessments. The amount o f capital that a bank
will achieve the appropriate debt rating. severity o f an event. Most common approach is must hold is specific to the risk o f credit-risky
• Give individual managers the information and the
lognormal distribution. assets, the type o f institution the claim is written
incentive they need to make decisions appropriate
Convolution-. Monte Carlo simulation combines on, and the maturity o f those assets.
to maintain the risk/capital tradeoff.
frequency and severity distributions. The internal ratings-based (IRB) approaches
The implementation steps o f ERM are as follows:
Rating Model Validation (foundation and advanced) use a bank’s own
• Identify the risks of the firm.
• Develop a consistent method to evaluate the firm’s Q ualitative validation-. (1) obtaining probabilities internal estimates o f creditworthiness to determine
exposure to the identified risks. o f default, (2) completeness, (3) objectivity, the risk weightings in the capital calculation.
• Foundation approach: bank estimates probability of
Firm-Wide VaR (4) acceptance, and (5) consistency.
default (PD).
• Firms that use value at risk (VaR) to assess potential Quantitative validation: (1) sample
• Advanced approach: bank estimates not only PD,
loss amounts will have multiple VaR measures to representativeness, (2) discriminatory power,
but also loss given default (LGD), exposure at
manage. (3) dynamic properties, and (4) calibration. default (EAD), and effective maturity (M).
• Market risk, credit risk, and operational risk will
Repurchase Agreements (Repos)
each produce its own VaR measures.
• Bilateral contracts where one party sells a security
• Due to diversification effects, firm-wide VaR will be
at a specified price with a commitment to buy back
less than the sum of the VaRs from each risk category.
the security at a future date at a higher price.
Market Risk Capital Requirements • Institute policies to address systemic risk and Momentum effect, long winners and short losers
Standardized method: determines capital charges interconnectedness. (W M L or winners minus losers). This strategy has
associated with various market risk exposures • Institute global liquidity standard (liquidity, outperformed both size and value/growth effects;
funding, and monitoring metrics). however, it is subject to crashes.
(equity risk, interest rate risk, foreign exchange
risk, commodity risk, and option risk). The Liquidity Coverage Ratio (LCR) Fundamental Law of Active
market risk capital charge for each market risk is Goal: ensure banks have adequate, high-quality
liquid assets to survive short-term stress scenario.
Management
computed as 8% o f its market-risky assets.
Tradeoff between required degree o f forecasting
Internal models approach (IMA): allows a bank LCR = (stock of high-quality liquid assets / total net
accuracy [information coefficient (IC)] and
to use its own risk management systems to cash outflows over next 30 calendar days) > 1 0 0
number o f investment bets placed [breadth (BR)].
determine its market risk capital charge. The Net Stable Funding Ratio (NSFR) IR stands for information ratio.
market risk charge is the higher o f (1) the Goal: protect banks over a longer time horizon
previous day’s VaR or (2) the average VaR over the than LCR. IR « I C x V B R
last 60 business days adjusted by a multiplicative NSFR = (available amount of stable funding /
factor (subject to a floor o f 3). required amount of stable funding) > 1 0 0 Illiquid Asset Return Biases
Biases that impact reported illiquid asset returns:
Backtesting VaR Stressed Value at Risk (SVaR) • Survivorship bias: Poor performing funds often quit
An exception occurs if the day’s change in value SVaR is calculated by combining current portfolio reporting results, ultimately fail, or never begin
exceeded the VaR estimate o f the previous day. performance data with the firm’s historical data reporting returns because performance is weak.
When backtesting VaR, the number o f exceptions from a significantly financial stressed period in the • Selection bias: Asset values and returns tend to be
is determined for a 250-day testing period. Based same portfolio. Calculation o f SVaR is defined as reported when they are high.
on the number o f exceptions, the bank’s exposure follows: • Infrequent trading: Betas, volatilities, and
is categorized into one o f three zones and VaR is max (SVaRt_ j, multiplicative factor x SVaRaVg) correlations are too low when they are computed
scaled up by the appropriate multiplier (subject to using the reported returns of infrequently traded
a floor o f 3). Solvency II assets.
• Green zone: 0—4 exceptions, increase in exposure Establishes capital requirements for the operational, Portfolio Construction Techniques
multiplier is 0. investment, and underwriting risks o f insurance • Screens simply choose assets by ranking alpha.
• Yellow zone: 5—9 exceptions, exposure multiplier companies. • Stratification chooses stocks based on screens;
increases between 0.4 and 0.85. • Specifies minimum capital requirements (MCR) includes assets from all asset classes.
• Red zone: Greater than or equal to 10 exceptions, and solvency capital requirements (SCR). • Linear programming attempts to construct a
multiplier increases by 1. • SCR may be calculated using either standardized portfolio that closely resembles the benchmark.
Operational Risk Capital approach or internal models approach. • Quadratic programming explicitly considers alpha,
• Standardized approach: Intended for less risk, and transactions costs.
Requirements sophisticated insurance firms; captures the risk
Basic indicator approach: measures the capital Portfolio Risk
profile of the average insurance firm.
charge on a firm-wide basis. Banks will hold capital Diversified VaR:
• Internal models approach: Similar to the IRB
for operational risk equal to a fixed percentage approach under Basel II. A VaR is calculated with a
VaRp = Z c X P x l w' a ' + +
o f the bank’s average annual gross income over one-year time horizon and a 99.5% confidence level. \ 2wj w 2CT! CT2Pi ;2
the prior three years. The Basel Committee has
Standardized Measurement Approach
proposed a fixed percentage equal to 15%. Undiversified VaR:
Standardized approach: allows banks to divide (SMA)
The SM A for operational risk includes both a VaRp = ^ V aR f + VaR^ + IV a R ^ a R ;,
activities along standardized business lines. = VaR! + VaR 2
Within each business line, gross income will be business indicator (BI) component accounting
multiplied by a fixed beta factor. The capital for operational risk exposure and an internal loss
VaRfo r Uncorrelated Positions:
charge for operational risk is the sum o f each multiplier (and loss component) accounting for
operational losses unique to an individual bank. VaRp = y jv aR? + VaR^
business line’s charges. The beta factors for the
eight business lines are as follows: The BI component impact will vary depending on
M arginal VaR: per dollar change in portfolio VaR
• Trading and sales: 18% where the bank is classified from buckets 1-5.
that occurs from an additional investment in a
• Corporate finance: 18%
position.
RISK MANAGEMENT AND
• Payment, settlement: 18%
• Commercial banking: 15% MVaRj = ---------------- x p.

INVESTMENT MANAGEMENT
• Agency services: 15% portfolio value
• Retail banking: 12%
• Retail brokerage: 12% How to use MVaR:
• Asset management: 12%
Factor Risks • Obtain the optimal portfolio: equate the excess
Represent exposures to bad times; must be return/MVaR ratios of all portfolio positions.
Advanced Measurement Approach (AMA): If
compensated for with risk premiums. Factor risk • Obtain the lowest portfolio VaR: equate just the
a bank can meet more rigorous supervisory
principles: MVaRs of all portfolio positions.
standards, it may use the AMA for operational
• It is not exposure to the specific asset that matters, Incremental VaR: change in VaR from the addition
risk capital calculations. The capital charge for
rather the exposure to the underlying risk factors. o f a new position in a portfolio.
AMA is calculated as the bank’s operational value • Assets represent bundles of factors, and assets’ risk Component VaR: amount o f risk a particular fund
at risk (OpVaR) with a 1-year horizon and a premiums reflect these risk factors. contributes to a portfolio o f funds.
99.9% confidence level. Having insurance can • Investors have different optimal exposures to risk CVaRj = MVaRj x wj x P = VaR x (3; x wj
reduce this capital charge by as much as 20% . factors, including volatility.
Basel III Changes Fama-French Model Risk Budgeting
• Raise capital standards (both quality and quantity). Explains asset returns based on: Manager establishes a risk budget for the entire
• Strengthen risk coverage of capital framework. • Traditional capital asset pricing model (CAPM) portfolio and then allocates risk to individual
• Require leverage ratio to supplement capital market risk factor.
requirements. positions based on a predetermined fund risk
• Factor that captures size effect (SMB or small cap
• Promote countercyclical buffers during financial level. The risk budgeting process differs from
minus big cap).
shocks. market value allocation since it involves the
• Factor that captures value/growth effect (HML or
high book-to-market value minus low book-to- allocation o f risk.
market value).
Budgeting risk across asset classes: selecting assets
whose combined VaRs are less than the total
The M -squared (M2) measure compares return
earned on the managed portfolio against the
CURRENT ISSUES IN FINANCIAL
allowed. market return, after adjusting for differences in MARKETS
Budgeting risk across active managers: the optimal standard deviations between the two portfolios.
allocation is achieved with the following formula: It can be illustrated by comparing the C M L for Cyber Risk and Market Stability
the market index and the CAL for the managed Systemic cyber risk is expected to increase the risk
weight of portfolio managed by manager i
portfolio. The difference in return between the o f financial system failures and instability over
IR i X portfolio’s tracking error time. Measures to improve resiliency to cyber
two portfolios equals the M 2 measure.
IR P X manager’ s tracking error risk include (1) approving all installed software,
Performance Attribution (2) standardizing secure system configurations,
Liquidity Duration Asset allocation attribution equals the difference
(3) having procedures to patch system
Approximation o f the number o f days necessary in returns attributable to active asset allocation
vulnerabilities, and (4) limiting administrative
to dispose o f a portfolio’s holdings without a decisions o f the portfolio manager.
access to the system.
significant market impact. Selection attribution equals the difference in
returns attributable to superior individual Big Data
T^ number of shares of a security
I. /I / security selection (correct selection o f mispriced Large datasets require tools that are more advanced
[desired max daily volume (%)
securities) and sector allocation (correct over- and than simple spreadsheet analysis. Overfitting and
X daily volume] variable selection are ongoing challenges. Tools
underweighting o f sectors within asset classes).
for analyzing big datasets include (1) classification
Time-Weighted and Dollar-Weighted Hedge Fund Strategies and regression trees, (2) cross-validation,
Returns Equity long/short strategy: go long and short
(3) conditional inference trees, (4) random forests,
Dollar-weighted rate o f return: the internal rate o f similar securities to exploit mispricings—
and (3) penalized regression.
return (IRR) on a portfolio taking into account decreases market risk and generates alpha.
Global macro strategy: makes leveraged bets on Machine Learning (ML)
all cash inflows and outflows.
anticipated price movements in broad equity Uses algorithms that allow computers to learn
Time-weighted rate o f return: measures compound
and fixed-income markets, interest rates, foreign without programming. Supervised M L predicts
growth. It is the rate at which $1 compounds over
exchange, and commodities. outcomes based on specific inputs, whereas
a specified time horizon.
M anagedfutures strategy: focuses on investments unsupervised M L analyzes data to identify patterns
Measures of Performance in bond, equity, commodity futures, and currency without estimating a dependent variable.
The Sharpe ratio calculates the amount o f excess M L can be applied to three classes o f statistical
markets around the world. Employs a high degree
return (over the risk-free rate) earned per unit problems: (1) regression, (2) classification, and
o f leverage because futures contracts are used.
o f total risk. It uses standard deviation as the (3) clustering.
Fixed-income arbitrage strategy: long/short strategy
relevant measure o f risk.
that looks for pricing inefficiencies between Artificial Intelligence (AI)
c _ Ra - Rf various fixed-income securities. The growing use o f Fintech, including AI and
-------------
aA Convertible arbitrage strategy: investor purchases M L, is driven by cost savings and increased
a convertible bond and sells short the underlying revenue. Firms use AI and M L to evaluate credit
where: stock. quality, optimize capital allocation, assess trading
= average account return Merger arbitrage strategy: involves purchasing impacts, predict changes in securities prices and
= average risk-free return shares in a target firm and selling short shares in price volatility, and perform regulatory functions
ct a = standard deviation o f account returns the purchasing firm. in financial markets.
The Treynor measure is very similar to the Sharpe Distressed investing strategy: purchase bonds o f Fintech
ratio except that it uses beta (systematic risk) as the distressed company and sell short the stock, Fintech brings technological innovation to
measure o f risk. It shows excess return (over the anticipating that the shares will eventually be the financial services industry. In addition to
risk-free rate) earned per unit o f systematic risk. worthless. the financial markets, Fintech is impacting
Ra - Rf Emerging markets strategy: invests in developing operations management (e.g., depersonalizing
1 a — ---------------
countries’ securities or sovereign debt. client interactions), leading and deposit services
Pa
Fund o f hedge funds: perform screening and due (e.g., automating credit scores), payment
where:
diligence o f other funds. Fees can be extensive, settlements, blockchain applications, and cross-
PA = average beta
and the due diligence does not always identify border payment services.
Jensens alpha is the difference between actual
fraud. A key advantage is diversification benefit
return and return required to compensate for
without large capital commitment.
Central Clearing
systematic risk. To calculate the measure, subtract Essentially eliminates the counterparty risk
the return calculated by the capital asset pricing Low-Risk Anomaly inherent in bilateral transactions by making the
Stocks with higher risk, measured by high standard
model (CAPM) from the account return. C C P the counterparty to each side o f the trade so
deviation or high beta, produce lower risk-adjusted
“ a = R a - E (r a) that virtually no default risk remains.
returns than stocks with lower risk.
Advantages o f central clearing include (1) halting
where: Explanation-, data mining, leverage and manager
constraints, and investor preferences. a potential domino effect o f defaults in a market
aA = alpha
downturn, (2) more clarity regarding the need for
E(Ra) = R„. +0 a [ E ( R J - R r]
collateral, (3) lower operational risk, (4) better
The information ratio is the ratio o f surplus return price discovery, (5) more regulatory transparency
(in a particular period) to its standard deviation. in O T C markets, and (6) better risk management.
ISBN: 978-1-4754-8481-6
It indicates the amount o f risk undertaken
Secured Overnight Financing Rate
(denominator) to achieve a certain level o f return
above the benchmark (numerator). (SOFR)
The SO F R is derived from a broad universe
td _ Ra - Rb
1r a ------------- o f actual overnight Treasury repo transactions.
cta - b
After applying filters to collected tri-party and
where: bilateral repo transactions, the rates for each day
a A. - B_ = standard deviation o f excess returns are weighted by transaction volume. The Chicago
measured as the difference between Mercantile Exchange (CM E) trades 1-month and
account and benchmark returns U.S. $29.00 © 2019 Kaplan, Inc. All Rights Reserved. 3-month SO F R futures.