Risk Factors and Risk Results
Asset Class, Risk Type and Sensitivities
The terms Asset Class, Instruments, Risk Types/Factors and Sensitivities are interpreted arbitrarily across the bank. Depending on the department dealing with Risk and its terminology, Risk jargon can change into a plethora of misleading concepts.
The aim of this page is to normalize the meaning of each Risk object contributing to calculate Greeks and VaR.
Asset Class
Asset Class is defined as the underlying element the Sensitivities are generated for.
Currently there are 10 Asset Classes modeled
Interest Rates | Basis | Credit | FX | Equity |
Commodity | Energy | Mortgage | Inflation | Other |
Temperature | Cross Risk |
Each Asset Class generates a number of Risk Types named with unique identifiers.
Asset Class | Risk Type | Risk Factor Description | VAR/ RNIV | Comments |
INTEREST RATE | IRDELTA | For all developed markets except the US, IR delta risk is modelled through the swap curve, and the basis risk between bond-like and swap-like instruments is measured as swap spread risk. For the US market, IR delta risk is modelled independently for swap and bond instruments. This is because the market practice in the US is that Treasury bonds are the benchmark for bond-like instruments, while in the non-US markets the swap curve is the benchmark for all fixed income instruments. For emerging markets, the most liquid curves in the market are used to calculate IR risk; these curves may be the government bond curve or the swap curve. All basis risks such as swap spread risk, currency basis risk and repo-spread risk are then measured over the respective reference curve. | VAR | |
INTEREST RATE | IRGAMMA | Second order interest rate sensitivity. The dollar value of the change in IRDELTA for a one basis point increase in the zero coupon rate. | VAR | |
INTEREST RATE | IRPCREVAL | The IR Reval risk for term structures is based on the revaluation with respect to the two leading principal components plus a delta correction to account for the yield curve movements that are not explained by these components. | VAR | |
INTEREST RATE | IRPCRESIDREVAL | Residual IR sensitivity | VAR | |
INTEREST RATE | IRSPOTVOLREVAL | 2D surface to capture cross effects between spot and volatility moves for interest rate products. | RNIV | Captured in primary and secondary framework where full revaluation is applied (e.g. for IR exotic positions) |
INTEREST RATE | IRVEGA | Sensitivity to interest rate volatility. The dollar value of a 1% absolute increase in the volatility of the underlying zero rate. | VAR | |
INTEREST RATE | IRSKEW | Impact of volatility skew (when slope of implied volatility against strike price is downward moving) and volatility smile (when slope of implied volatility against strike price is upward moving). Measured with the SABR (Stochastic Alpha, Beta, Rho) methodology where Stochastic Alpha, Beta, Rho are the input parameters of the model. | RNIV | |
INTEREST RATE | IRTENORREVAL | The majority of government bond options can be revalued individually for changes in their yield. These are bucketed by their duration (linearly interpolated into the two nearest buckets if necessary). In the non-US markets the changes are simulated against this single tenor of the swap curve or the relevant single government bond yield data – if available. In the US market they are simulated against the bond curve or against the single government bond yield data – if available. | VAR | |
BASIS | CCYBASISSWAP | The exchange of two floating rate interest streams of different currencies and of different payment basis (either payment frequency, day count or business day convention). For each currency, the DV01 is calculated as the P&L impact for an actual 1 bp increase of the currency basis swap spread level. | VAR | |
BASIS | INTRACCYBASIS | Defines the basis risk. | RNIV | |
BASIS | NONLIBOR | The basis spread risk type is used to measure the basis risk between various traded rates e.g. a non-Libor interest rate and the comparable Libor (or reference) rate. Examples of basis spreads are spreads between the Japanese Tibor rates, the French TAM rates, the US CP, Fed Fund, Prime OIS and PSA rates and the relevant reference curves. Also, some swap rates on bond indices such as the German REX10CMS and the French TEC10CMS are modelled as spreads over the respective CMS rates. These are therefore also classified as basis spread risks and treated in the same way as described in this section, and not as separate bond indices. Note that some basis spreads such as repo spreads are classified as a risk type in their own right. Basis spread risk is a first order market risk, and the risk calculation is done by the simulation method. Some, but not all, basis spreads are modelled as a term-structure. The term structure of the basis spread risk corresponds to the term structure of the base rate. Examples of basis spreads with a term structure are US agency spreads. | VAR | Specific scenarios run for LIBOR-OIS spread risk as part of secondary framework |
BASIS | SWAPSPREAD | Swap spread risk measures the basis risk between swap-like instruments (money market positions, interest rate futures, swaps, caps/floors, swaptions etc.) and government bond-like instruments (bonds, bond futures, bond options etc). Swap spread risk is a first order market risk, and the risk calculation is done by the simulation method. Swap spread risk is calculated by currency and for term structures. For term structures that cannot be simulated, the simple sum method is used to aggregate the term-structure. Swap spread risk is only relevant for markets for which the swap and bond delta risk is not modelled independently. Currently these are all markets except USD. Note that swap spread risk is only measured for government bonds which are issued in their local currency and are hedged against swaps. Swap spread risk is not measured for corporate bonds as for these the credit spreads are measured against the reference curve. Similarly, as government bonds issued in a foreign currency are treated like corporate bonds in this currency, for these bonds no swap spread risk is measured. As the volatilities of swap rates and bond yields of the same maturity are very similar and their correlations are very high, swap spread risk is not applied to outright bond positions (or in emerging markets, where the reference curve is the bond curve, outright swap positions); their risk is adequately measured via the reference curve. Swap spread risk is only measured for offsetting swap and bond positions, for which the interest rate delta DV01s are netted (and zero). | VAR | Captured in primary swap spread scenarios |
BASIS | INTRAMATSPREAD | Intra-maturity spread risk captures the risk due to price movements between two different government bond instruments in the same market and with a similar maturity. Note that spread risk between two bond positions in different markets is implicitly captured by simultaneously historically simulating these positions through the different bond curves. Similarly, spread risk between two bond positions in the same market but with different maturities is implicitly captured by simultaneously simulating these positions through the different maturity buckets. However, simultaneous simulation will not capture the risk when two different bonds from the same market with similar maturities trade at a small spread (mainly due to liquidity and investor preferences). To capture this risk the 10-day spread moves of bonds within the same maturity bucket (and from the same market) are measured. For other holding periods the same scaling rule is applied. Intra-maturity spread risk is a first order market risk. The risk calculation for this risk-type is done by the zero-correlation method, i.e. the risk calculation is entirely extreme move based and it is assumed that the different spreads are uncorrelated with each other and with any other risk (except for the Euro cross-market terms). Intra-maturity spread risk is calculated only for the major markets (currently USD, EUR and JPY) and is only measured for government bonds issued in their local currency. | VAR | |
BASIS | CASHFUTUREBASIS | Measures the impact of difference in price changes of cash bonds and bond futures. Conservatively estimated as the simple sum across all currencies (widening or narrowing occurs simultaneously in all currencies). | RNIV | |
CREDIT | CREDSPRDCORPBOND | Single corporate bond credit spread risk is a first order specific risk. The risk calculation for this risk type is done by the zero-correlation method, i.e. the risk calculation is entirely extreme move based and it is assumed that the specific risks of different bond issuers are uncorrelated with each other and with any other risk. Single corporate bond credit spread risk is not calculated for each individual issue, but for each issuer. The risk arising from different issues from the same issuer is simply summed, i.e. a correlation of 100% is used. The credit spread risk for corporate bonds is calculated via a single factor model. The risk is split into a market risk part and a specific risk part. The market risk part is treated together with any other market risks arising from other corporate bonds and index positions, and is simulated whenever possible. It is assumed that for every corporate bond the market risk spread has been calculated separately and is labelled as such. This section deals only with the specific risk calculation. | VAR | Primary scenario reporting includes a general spread widening by credit rating rather than an issuer specific event scenario, this is supplemented with a range of cluster scenarios |
CREDIT | CREDSPRDCORPBONDINDEX | Corporate bond index credit spread risk is a first order market risk, which is included in the ASR (Additional Specific Risk). The risk calculation for this risk type is done by the simulation method. Corporate bond index credit spread risk is calculated for every G21 corporate bond, foreign currency G21 government bonds and for EM corporate bonds issued in local currency. In addition, single corporate bond credit spread risk is calculated for every corporate bond, which is the specific risk of these bonds. It is assumed that for each corporate bond two credit spread sensitivities are generated: one sensitivity for the market risk (by country, rating class and maturity), and one sensitivity for the specific risk (by country, rating class and issuer). Note that the term structure used for all currencies is extended at the short end for USD to accommodate for short-dated commercial paper (for the other currencies these maturities do not exist). The market (or systemic) risk of G21 and local currency EM corporate bonds is measured via synthetic corporate bond indices. These capture the moves of the generic credit spread curve for each rating class, as opposed the moves which are due to changes in the credit spreads of a particular corporate bond. The latter constitutes the specific risk of corporate bonds. Some G21 corporate bond positions, usually those that are part of a risk arbitrage position, are fully simulated and are not included in this risk type. These positions need to be flagged accordingly. | VAR | N/A for G22 as credit spread delta is used |
CREDIT | CREDSPRDDELTA | Credit spread delta risk is a first order specific risk. The credit spread risk for corporate bonds is calculated via a single factor regression model with specific risk. Sensitivities (CS01) are regressed on credit indices by region and credit type (cash and CDS.) Spread changes by corporate names are calculated as described above, and P&L strips are generated. To calculate specific risk, CS01 is aggregated by name and applied to a proportional XM. | RNIV | Captured in G22 primary credit spread scenario |
CREDIT | CREDSPRDEMBOND | Emerging market credit spread delta risk is a first order market risk, which is included in the ASR. The risk calculation for this risk type is done by the simulation method. Emerging market credit spread delta risk is calculated for every hard currency emerging market bond, i.e. for both sovereign and corporate bonds. For corporate bonds, single corporate bond credit spread risk is calculated in addition, which is the specific risk of these bonds. It is assumed that for each corporate bond two credit spread sensitivities are generated: one for market risk and one for specific risk (by country, rating class and issuer). As the credit spread behaviour of different bonds issued by the same country can be quite different (compare Brady bonds to Euro bonds), separate curves might be used in this instance. Currently, separate curves are used for Brazil (an on-shore and an off-shore curve). Some EM bond positions, usually those that are part of a risk arbitrage position, are fully simulated and are not included in this risk type. These positions need to be flagged accordingly. | VAR | Captured in EM Spread scenario |
CREDIT | TRADED LOAN DELTA | The traded loans VaR model is a one factor regression model in which each loan is mapped to its corresponding sector index. The traded loans delta risk type captures both market and specific risk. The loans’ beta coefficients are calculated by risk rating and sector by regressing the loans data on the sector indices and used for the market risk calculation. Specific risk is captured by an extreme moves approach. | VAR | Captured in primary credit spread scenario |
CREDIT | CREDSPRDREVAL | Revaluation surface for credit spread moves. | RNIV | Risk type in development for scenario use, however partially covered by CREDSPRDDELTA |
CREDIT | HIGHYIELDDELTA | High yield risk sensitivity. Split into index and specific risk during calculation. | VAR | Captured in primary credit spread scenario |
CREDIT | CROSS GAMMA | Second order cross risk is defined as the cross gamma risk arising from a simultaneous change in two underlying risk factors. Cross gamma is a quadratic risk (hence the name second order cross risk). For example, for IR versus IR cross-gamma, the sensitivity needed is the change in the DV01 (of the first currency) for an actual 1bp increase in the zero rates of the second currency. This means that the IR-IR cross-gamma is equal to the corresponding mathematical sensitivity divided by 100,000,000 (to adjust for an actual 1bp move). | RNIV | Where full revaluation is applied some cross gamma is captured in the scenario, however cross gamma effects have to be considered on a case by case basis |
CREDIT | TRADEDLOANGAMMA | Gamma sensitivity for traded loans (same underlying regression models as for traded loans delta). | RNIV | Traded Loan gamma is not explicitly captured in the scenario framework. However, price shocks are applied to leveraged loan delta in the primary scenario framework which would capture this risk on a vanilla loan position |
CREDIT | RECOVERY SENSITIVITY | Sensitivity to changes in recovery assumptions. | RNIV | |
CREDIT | CREDIT SPREAD SKEW | Risk between CDS index and individual names. | RNIV | Captured in primary traded credit scenarios by applying different shocks to single name CDS and index CDS positions |
CREDIT | CREDIT SPREAD VEGA | Impact of changes in volatility on option’s credit spread. | RNIV | Capturing optionality on credit positions is currently under review |
FX | FXDELTA | First order FX sensitivity. The dollar amount of currency or equivalent that is being held. | VAR | |
FX | FXGAMMA | Second order FX sensitivity. The dollar value of the change in the position for a 1% proportional increase in the appropriate spot FX rate. | VAR | |
FX | FXREVAL | Partial revaluation surface for non-linear FX positions. | VAR | |
FX | FXSPOTVOLCROSSRISK | 2D surface to capture cross effects between spot and volatility moves for FX products. | RNIV | Captured in primary and secondary framework where full revaluation is applied |
FX | FXVEGA | Sensitivity to FX volatility. The dollar value of a 1% absolute increase in the appropriate FX rate implied volatility. | VAR | |
FX | FXSKEW | Impact of FX volatility skew (risk reversal) and smile (strangle premium). | RNIV | Captured in exotic business specific scenarios |
EQUITY | EQDELTASTOCK | Some single stock positions (for example those that are part of a risk arbitrage deal), are best modelled individually and are fully simulated. These positions need to be flagged accordingly (as the default is inclusion in the regression model). All risk stemming from a full simulation of a single stock position is classified as specific risk (note however that if the risk arbitrage position is set up correctly, then there should be no market risk, and the simulated risk will reflect this). Different positions in the same stock could be treated in the two different models. | VAR | |
EQUITY | EQGAMMASTOCK | Second order equity sensitivity. | VAR | |
EQUITY | EQSURFACESTOCK | Due to the approximate nature of the regression model, large stock option positions are best modelled individually and are fully simulated. These positions need to be flagged accordingly (as the default is inclusion in the regression model). All risk stemming from a full simulation of a stock option position is currently classified as specific risk; this however may be reviewed at a later stage. Different option positions in the same stock should be treated under one model only (regression or simulation). | VAR | |
EQUITY | EQ SPOT VOL REVAL | 2D surface to capture cross effects between spot and volatility moves for equity products. | RNIV | Captured in primary and secondary framework where full revaluation is applied |
EQUITY | EQVEGA | Sensitivity to equity price volatility. The dollar value of a 1% absolute increase in the volatility of the underlying equity (index) price (needs clarification between stock and index prices). | VAR | |
EQUITY | EQ VOLSKEW | Impact of volatility skew steepening or flattening. | RNIV | Captured in vol skew steepening scenario |
EQUITY | EQ BORROW COST | Measures the impact of changes in expected borrow costs on forward prices (and hence option portfolio). | RNIV | J. Louis to review |
EQUITY | DIVIDENDRISK | Measures the impact of changes in expected dividend income on forward prices (and hence option portfolio). | RNIV | |
COMMODITY | COMDELTA | Commodity delta risk refers to first order commodity risk arising from any commodity other than metals or energy. Commodity exposure is quoted in terms of the different futures contracts. As such commodity delta risk is measured via a term structure (rather than as spot delta risk plus a lease rate curve). Commodity delta risk is a first order market risk, and the risk calculation is done by the simulation method. The risk is calculated by commodity. For term structures that cannot be simulated, the simple sum method is used to aggregate the term structure. | VAR | |
COMMODITY | COMLEASEDELTA | First order commodity lease sensitivity. The P&L impact for a 1bp increase of the zero rate for the commodity lease. | VAR | |
COMMODITY | COMGAMMA | Second order commodity price sensitivity. The dollar value of the change in the commodity position for proportional 1% increase in the underlying commodity price. | VAR | |
COMMODITY | COMREVAL | Partial revaluation surface for commodity positions. | RNIV | Full revaluation scenarios currently being developed for commodities |
COMMODITY | COMVEGA | Sensitivity to commodity price volatility. The dollar value of a 1% absolute increase in the volatility of the underlying commodity price. | VAR | Not captured in scenarios – volatility shocks currently being calibrated |
COMMODITY | COMSKEW | Sensitivity to volatility skew for commodity derivatives. | RNIV | |
COMMODITY | COM SPOT VOL REVAL | 2D surface to capture cross effects between spot and volatility moves for commodity products. | RNIV | Will be captured via full revaluation once volatility shocks are defined |
COMMODITY | COM-FX CROSSRISK | Cross gamma between commodity and FX. | RNIV | |
ENERGY | ENERGYDELTA | Energy delta risk is measured via a term structure (rather than as spot delta risk plus a lease rate curve). Energy delta risk is a first order market risk, and the risk calculation is done by the simulation method. The risk is calculated by energy type. | VAR | |
ENERGY | ENERGYGAMMA | For each energy position, the gamma is calculated as the change of the position (i.e. the delta) for a proportional 1% increase in the underlying price, expressed in USD. | VAR | |
ENERGY | ENERGYREVAL | The energy reval risk for term structures is based on the revaluation with respect to the two leading principal components plus a delta correction to account for the term structure movements that are not explained by these components. | VAR | |
ENERGY | ENERGYSPREADREVAL | Non-linear sensitivity to energy spread. | VAR | |
ENERGY | ENERGY CROSS GAMMA | Cross gamma between two different energy types (or tenor points on one curve). | RNIV | |
ENERGY | ENERGY SPOT VOL REVAL | 2D surface to capture cross effects between spot and volatility moves for energy products. | RNIV | |
ENERGY | ENERGYVEGA | Energy Vega is calculated as the P&L impact for an actual 1% increase of the respective volatility. | VAR | Not captured in scenarios – volatility shocks currently being calibrated |
MORTGAGE | OASDELTA | Mortgage spread risk represents the risk of agency mortgage securities to the movements of the Option Adjusted Spread (OAS). The DV01 (no term structure) is calculated as the P&L impact for an actual 1 bp parallel increase of the OAS. | VAR | Primary scenarios apply price based shocks to mortgage positions and do not use the OAS sensitivities |
MORTGAGE | OASDELTASPECIFIC | Specific (idiosyncratic) component of OAS sensitivity for structured products. | VAR | Primary scenarios apply price based shocks to mortgage positions and do not use the OAS sensitivities |
MORTGAGE | MORTVEGA | This risk type is used to evaluate the risk to the movements of interest rate volatility. All securities are mapped to one of the benchmark securities based on the security characteristics. Each security mapped to a benchmark security will have a scaling factor associated with it. This scaling factor represents the ratio of parallel shift Vegas between the specific security and the benchmark security. | VAR | Full revaluation scenarios are being prioritised for development within the Structured Products cluster |
INFLATION | INFLATIONDELTA | Linear sensitivity to inflation risk. | RNIV | Captured in exotic inflation scenarios via full revaluation |
INFLATION | INFLATIONGAMMA | Second order inflation sensitivity to inflation risk. | RNIV | Captured in exotic inflation scenarios via full revaluation |
INFLATION | INFLATION REVAL | Non-linear sensitivity to inflation risk. | RNIV | Captured in exotic inflation scenarios via full revaluation |
INFLATION | INFLATIONVEGA | Sensitivity to inflation volatility. | RNIV | Captured in exotic inflation scenarios via full revaluation |
OTHER | CORRELATIONDELTA | Correlation delta risk is defined as the risk arising from a change in correlation between two risk factors. Correlation delta is a linear risk, which can be approximated by a linear move of the correlation. The sensitivity needed is the P&L impact for an actual 1% increase of the respective correlation. This means that the correlation sensitivity is equal to the corresponding mathematical sensitivity scaled by 1/100. | VAR | Correlation scenarios are applied to the equity portfolios which are included in equity crash, and correlation scenarios are run for Exotics desk as part of the business specific scenario framework |
OTHER | CROSSRISKSECOND | Second order cross risk is defined as the cross-gamma risk arising from a simultaneous change in two underlying risk factors. Cross-gamma is a quadratic risk (hence the name second order cross risk). Some options (such as quanto options or spread options) depend on two underlying risk factors simultaneously, that is the change in value (or P&L) is not the sum of the change in value implied by the moves in both risk factors individually. The Taylor approximation used to estimate the P&L impact from the underlying historical moves needs to be expanded to include the second order cross terms. The first four terms in the approximation are straight delta and gamma terms, the last term is the cross gamma term, the portfolio’s sensitivity with respect to a simultaneous move in both risk factors. | VAR | Cross gamma effects need to be considered on a case by case basis. Cross gamma effects cannot be easily segregated in scenario results due to the application of simultaneous shocks |
OTHER | EVENT RISK | The risk of events that have not necessarily happened to the specific security under consideration. The strategy is to implement event risk selectively where we have explicit and significant event risk (e.g. merger arbitrage positions). | RNIV | Risk arb deal probability break up is included within Equity Crash scenarios. FLP scenarios capture gap risk |
OTHER | CONSTDEFRATE | Annualized default rate on a pool of underlings. | RNIV | |
OTHER | CONSTPREPYMTRATE | Annualized percentage of outstanding mortgage loan principal that prepays in one year. | RNIV |
Note: Some Risk Factors may derive from particular trading strategies and are termed Risk Not In VaR (RNIV). Risk Sensitivities are generally associated to Risk Factors, but there are cases where this relationship may not occur.
Risk Factors
Risk Factors are the underlying market (and credit) risk variables driving the trade’s VM and Sensitivities measures. These measures represents the different characteristics of risk that an Risk Class (or type) may exhibit, such as Sensitivity to changes in price or interest rates. The measures are academically known as Greeks (Delta, Gamma, Theta, Rho, and Vega, etc).
Risk Factors include: IR, FX Rates, Equity Prices, Commodity Prices, Credit Spreads, Recovery Rates, Implied correlation, Historical or implied volatilities
The identification of market risk factors in the existing business is performed using a “bottom-up” methodology whereby risk factors are identified based on the asset type. (i.e. for each traded asset type, a set of risk factors that drive the underlying pricing model is defined). Market Risk Factors determine what market data is required to price a trade and therefore what risk sensitivities are relevant for that trade.
Risk Factors/Sensitivities Example:
FX Options trades
Asset Class: ‘FX’,
Risk Factors:
- Buy Ccy IR
- Sell Ccy IR
- FX rate
- Implied Vol
Sensitivities:
- FXDELTA
- FXGAMMA
- FXREVAL
- FXSPOTVOLCROSSRISK
- FXVEGA
- FXSKEW
Technically, a risk factor is associated with a particular type of perturbation to an object. Perturbation may impact the equity spot value or a vol surface parameter whilst general general perturbations may be represented by external skew shift to a volatility surface. Some Risk Factors can produce straightforward values such as the Equity Spot Value for an EQSpot Risk Factor. In this case, the Risk factor can be used in the RBPL (Risk Based P&L) report. Other Risk Factors may not be that intuitive, such as a parallel shift of a vol surface. In this case, an approximation or no value at all is provided.
Risk Factor keys
Market Risk is expressed in terms of risk factor “keys”. A risk factor key determines the object being perturbed + the type of perturbation. A risk factor key is a record type, akin to a market key.
The risk model will map a risk factor result type to one or more risk factor keys to determine the perturbations