Fixed Income Flashcards
Excess Spread
(define and calculate)
Spread in excess of the fair spread for suffering credit losses; an unannualized excess return over a specific holding period (NOTE: may need to deannualize to match hold period)
important to analyze source of mgr’s excess return
goal of any mgr is to earn excess return over fair mkt returns by finding mispriced securities
E [ExcessSpread] ≈ Spread0 − (EffSpreadDur × ΔSpread) − (POD × LGD)
***NOTE: the first term (Spread0) might need to be adjusted for periods in the year (eg 6 month hold would be x 0.5), AS DOES THE PODxLGD TERM
CDS Spread
(define and calculate)
standardized fixed coupon adjusted by an up-front pmt/receipt to freflect FV of the protection
essentially a premium paid to party who has credit risk of the bond
CDS Price ≈ 1 + (Fixed coupon − CDS spread) × EffSpreadDurCDS

Inflation-linked Bonds
Principal expands in line with an inflation index
exhibit lower return volatility than conventional bonds and equities
returns depend on volatility of REAL, not NOMINAL interest rates
–> volatility of real rates typically lower than that of nominal rates, which impact returns of conentional bonds and equities
Calculate Bond Portfolio Returns
given key rate durations and a change in yield curve, use the change in yield curve for the specified rate duration
Sum of Durations for each tenor times the change in yield
Modified Duration
% change in bond price for a 1% change in yield
Macaulay / (1+ periodic discount rate)
NOTE: make sure to divide by coupon frequency for periodic discount rate
if MD is 7, value of bond expected to decrease by 7% with a 1% increase in yield
Effective Duration
used for embedded options
HINT: think E for effective and E for embedded
if no options, effdur will be same as Modified Duration
Money Duration
(Modified Duration) x (change in yield) x (mkt value)
–> Multiply by 1/1002 to get BPV (price value of a basis point)
related to dollar sensitivity
Spread Duration
measures a credit-risky bond’s sensitivity to changes in credit spreads
Key point: NOT looking at underlying benchmark rates, just the spread movement
Higher effective spread duration = higher sensitivity to changes in spread
Empirical Duration
takes step back and looks at how prices move in actual terms (“empirical,” duh)
how muc hdoes bond price ACTUALLY move from shift in rates
just looks at regression of historical price and int rate changes
Return Decomposition
FS1, v important
- Yield income: coupon / price
-
Rolldown Return: return bond will have if the yield curve does not change
- think of upward sloping curve (maturity horizontal yield vertical axis)
- Rolling Yield = #1 + #2 (note difference, it’s NOT rolldown return)
- once you have 1&2 you can think about what is attributable to changes in benchmark rates and and spreads (ie changes in the yield curve), steps 3 and 4:
- Manager predicted change based on duration, convexity, and benchmark rates
- first component of this formula is the duration change, second is the “second order” change, due to shape:
- see photo
- mgr predicted price changed based on D, C, and yield SPREADS
- projected foreign currency G/L
- if foreign currency strengthens you benefit
- LESS expected credit losses

Structural Risk
the risk of non-parallel shifts in yield curve
causes assets to act differently from liabilities –> might not be able to meet liabilities when they come due
non-parallel shifts = diff rates move by diff amts
Convexity = bad from an immunization point of view –> dispersion –> higher dispersion = higher exposure to non-parallel yield shifts = higher structural risk
Immunizing Multiple Liabilities
PVA >= PVL
BPVa = BPVL (so assets move the same amount as liabilities with changes in yield curve)
convexity of assets slightly exceeds Convexity of liabilities –> look for the minimum convexity you can get because you DO need dispersion of asset cash flows to be slightly wider than that of liabilities because asset cash flow needs to happen before Liability cash flow
BPV
MD x V x 0.0001
Duration Gap
BPVA - BPVL
if <0, need to increase duration of assets
can do so by going long futures
of futures needed = desired change in BPV / BPVfutures (of one futures contract)
BPV of one futures contract = BPVCTD / CVCTD
conversion factor maps the CTD to a notional bond underlying the futures contract
similar with swaps:
Size(NP) = change in BPV / BPVswap
Butterfly Spread and Formula
butterfly trade = leveraged way to cpature value when curvature changes
take long and offsetting short position in the bullet and offsetting barbell
short funds the long so no investor capital is required
long and short duration CANCEL each other for 0 net duration
profit primarily from change in curvature
- (ST yield) + (2 x medium-term yield) - LT yield
Forward Rate Bias
defined as an observed divergence from IRP, where active investors seek to profit by borrowing in lower-yielding currency and investing in a higher-yielding curency
lower-yielding currencies trade at a forward PREMIUM
fully hedged foreing currency FI invmts will tend to yield the domestic RfR
G Spread
the bond’s YTM minus an interpolated YTM of the two adjacent maturity OTR gov bonds
estimates a gov benchmark YTM that exactly matches the maturity of the bond
YIELD SPREAD = just the corp bond’s yield minus the nearest dated gove bond
to get to G-spread need to interpolate:
(M - S) / (L-S) –> easier way, M = target in the middle –> gives the weight of the LONGER dated benchmark bond
target = (1-weight) x shorter + (weight x longer)
if gov bond yield curve is flat, it will equal the Yield Spread
(example p.103)
I-spread
bond’s YTM minus maturity interpolated swap fixed rate
advantage: based on a tradeable derivative that can be used to hedge inflation or measure carry returns
Calculate: if only given yield, first add the adjacent two swap spreads to their gov bond (benchmark) yields
then interpolate the same way as G-spread:
(M-S) / (L-S)
example p. 105
ASW
Asset Swap Spread
bond’s fixed coupon minus the maturity interpolated swap fixed rate
ie it’s the same as the G-spread but using the COUPON instead of the bond’s YTM to find the spread
NOTE: can do the same thing if given an i-spread…just figure out i spread first and swap out coupon for the swap rate (instead of swapping for YTM for g-spread)
Yield Spread
simple difference between bond’s all-in YTM and the closest maturity current OTR gov bond
if gov bond yield curve is flat, it will be equal to the g-spread
Structural vs. RF Credit Models
Structural:
use mkt-based variables to estimate issuer’s asset value and asset value volatility
define likelihood of default as probability of asset value falling below that of liabilities
zero net assets = default threshold
so “distance to default” (DtD) is the measure
Reduced-Form:
look for relations between macro conditions and borrowers
estimate POD or spreads:
POD ≈ Credit Spread / LGD
Altman’s Z-Score
used in RF model
uses key financial ratios weighted by coefficients - dont need to know the model, just need to be able to plug in
INTERPRETATION:
> 3.0 = low chance of default
1.8 - 3.0 = some risk of default
< 1.8 = likely to default
VaR Methods
Parametric
assumes returns normal
uses return and stdev to calculate loss at given percentile of distribution
5% = 1.65
1% = 2.33
Advantage: simple
disadvantage: doesnt work well for non-normal distributions (eg DOESNT work with options)
Historical Simulation
applies hist movements in key risk factors (rates spreads etc) to existing portfolio to generate return dist
advantage: uses actual mkt data, CAN handle portfolios with non-normal
Disadvantage: heavily dependent on historical sample, and history might not be guide for the future
Monte Carlo Analysis
generates return distr through random simulations from user defined model
advantage: flexilibty to incorporate non-normal distr
disadvantage: high model risk
Calculate VaR
1) scale up yield volatility by multiplying the square root of the time (in days) –> recall that variance scales up with time, so StDev (the sqrt of variance) scales up by multiplying by the SqRt of time)
2) multiply that by the yield to get to ONE standard deviation
3) multiply by # of StDevs (eg 2.33) –> gives you VaR
4) Multiply that by negative MD to get % change in value
5) multiply that by portfolio value to get $ change in value
Example in photo

Drawbacks of VaR and How to Fix them (ie cVaR etc)
VaR as a tail risk measure has the following drawbacks:
- Tail events tend to be more frequent and more severe than VaR forecasts.
- It fails to capture changes in correlation and liquidity during times of market stress.
- It fails to quantify the expected loss during a tail event.
Fixes:
Conditional value at risk (CVaR) = expected loss given portfolio is experiencing a loss in the tail –> a measure of the avg loss in the tail –> addresses the third drawback listed previously.
Incremental VaR (IVaR) (or partial VaR) measures change in VaR from adding or removing a position in a portfolio.
Relative VaR measures VaR portfolio’s returns relative to benchmark –> can be used to assess minimum underperformance of an active manager relative to their benchmark in a given time frame with a specified probability.
Convexity and Effective Convexity
convexity = second order measure of the non-linear effects of a yield change of bond price –> the rate of change of the rate of change
–> a good thing if we expect rates to change
–> longer dated bonds have higher convexity
* positive convexity = causes prices to rise by more and fall by less than duration predicts
crucial takeaway: for a given value of duration, the more spread out cash flows are, the higher convexity is
Effective Convexity = used for bonds w uncertain CFs (eg options)
Duration Times Spread (DTS)
adjusts spread duration to give a higher sensitivity to spread changes when spreads themselves are higher
–> because bonds w/ higher spreads have bigger spread changes
–> eg if spread is 10%, a 1% move isnt a big deal, but would be for a bond w a spread of 1%
DTS = (EffSpreadDur) x (Spread)
Stage of Economic Cycle and Effects on Defaults and Spreads

Tracking Error for Bond Portfolios
Tracking error = standard deviation of active return of a portfolio
if normally distributed, portfolio return should be within +/- 1 stddev of the index’s return in 68% of time periods
so if you build a portfolio mimicking an index with tracking error of 1%, with normal distribution you should expect to achieve a return that’s within 100 bps of the return on the index 68% of the time
Calculate Expected Change in Portfolio Price (based on expected yield and yield spread changes)
(-EffDur x change in yield) + (0.5 x effConvexity x yieldchange 2)
Key Rate Duration Formula
KeyRateDur = - (change in port value) / (port value x change in key rate)

Leverage to Enhance Port Return Formula
rp = rl + [(Vb / Ve) x (rl - rb)]
Calculate P/L of a CDS Position
Duration neutral heding position example from CFA Mock 1, # 17:
In order to be duration neutral, the BPV of the $1,000,000 notional value of protection sold on the 10-year index must match the BPV of the protection bought in the 3-year contract.
The BPV of the $1,000,000 notional value of protection sold on the 10-year index is equal to its effective spread duration multiplied by the notional value as follows:
BPV10-year = 9.65 × $1,000,000 = $9,650,000
To ensure that the BPV of the 3-year index is the same size, we require:
$9,650,000 = 2.85 × Notional3-year
This implies the appropriate notional value in the 3-year index = $9,650,000 / 2.85 = $3,385,965.
The absolute profit/loss on each contract is calculated as ∆CDS spread × Effective spread duration × Notional amount. Given that spreads are falling, there will be profits from selling protection in the 10-year contract and losses from buying protection in the 3-year contract.
Loss in the 3-year contract = (50 / 10,000) × 2.85 × $3,385,965 = $48,250
Profit in the 10-year contract = (30 / 10,000) × 9.65 × $1,000,000 = $28,950
Hence, net loss = –$48,250 + $28,950 = –$19,300.
Immunizing a Single Liability
1) Initial MV value equals or exceeds the PV of the liability
2) Macaulay duration that matches the due date (as close as possible)
3) minimizes portfolio convexity
Best Type of Portfolio for taking advantage of an expected FLATTENING yield curve
Barbell portfolio
–> the price impact of the longer-dated bond (which will go up in value bc yields going down) will be much larger than that of the shorter one (which will go down in value bc ST yields going up in flattening) because the duration of the longer dated bond(s) is larger than that of the shorter dated one(s)