Types of fixed income risk
This section provides an overview of the various types of fixed income return that can be measured by FIA.
Familiarity with terms such as bond, zero coupon yield curve, credit curve is assumed.
Table of effects
||Return generated by the passage of time. Even if yield curves do not move at all, these effects will continue to generate return:
is the security's coupon, if it has one;
is the security's clean price;
is the elapsed time as a fraction of a year;
is the risk-free portion of a security's yield;
is the overall security market yield, including both risk-free portion and the portion generated by creditworthiness effects, if applicable.
||Return generated by the slope of the yield curve|
|Sovereign curve return||
||Return generated by changes in the risk-free curve|
|Credit curve return||
||Return generated by changes in the spread between the risk-free curve and a sector or credit curve.|
||Return generated that is not accounted for elsewhere.|
||Return generated by securities for which no pricing type has been defined|
||Return generated by non-linear dependency of price on yield|
||Return generated by interest on cash deposits|
||Return generated by inflation for index-linked bonds|
|Asset allocation return||
||Return generated by over- or under-weighting market sectors|
|Duration and spread duration allocation return||
||Return generated by contribution to duration|
Asset allocation and stock selection return
This way of breaking down active portfolio returns against benchmark is usually associated with equity portfolios, but it can equally well be applied to fixed income portfolios.
Asset allocation return measures the return generated by under- or over-weighting the portfolio's exposures against those of the benchmark, where exposures are grouped into sector groups. Stock selection is the return made in each sector by selecting stocks. The sum of asset allocation and stock selection returns, plus interaction returns if applicable, equals the overall active return of the portfolio against the benchmark.
Use the BrinsonAllocationSectors field to assign the sector to be used in the calculation. If no sectors are assigned, no asset allocation attribution is performed.
Multiple sectors can be allocated to BrinsonAllocationSectors, allowing asset allocation returns to be calculated at multiple levels.
Asset allocation and stock selection returns are described in more detail elsewhere on this site.
Unlike equities, coupon-bearing securities guarantee a certain return to the holder due to the passage of time. If the markets did not move in any way, the holder of a coupon-bearing bond would still receive one or more coupon payments per year. If the holder only owned the bond for part of the year, he would receive part of the regular coupon payment, with the amount proportional to the time he had owned the security. The majority of fixed-income securities pay some sort of coupon, so this is an important source of return in the fixed income markets.
The terms coupon return and time return are often used interchangably, since all can be seen as names for return that does not arise from changes in the market's term structure. However, the two terms are not strictly equivalent. Securities that pay no coupon, such as zero-coupon bonds or bank bills, show return that is purely due to the passage of time, which causes the security's price to approach par. A better alternative is 'yield return' or 'carry'.
There are several ways in which one can measure a security's return due to the passage of time while ignoring changes in the term structure. One of the simplest is to use the security's yield to maturity (or YTM). YTM is the single rate that, when used to discount a security’s cash flows, gives the current security price :
where is cash flow , and is the interval in years between the present and that cash flow.
YTM has the property that, if unchanged over a short period, it equals the security’s total rate of return. A bond’s return due to yield is therefore given by
where is the security’s yield to maturity, and is the elapsed time in years.
Current yield and pull to par
Yield return may be further decomposed into a current yield and a pull-to-par yield.
Suppose a bond has just been issued and has a long time to maturity. To a close approximation, its instantaneous return will be given by its coupon divided by the price at which it was bought. This quantity is called the current yield (also flat yield, interest yield, or running yield), and is given by
where is the current yield return, is the bond’s coupon, and is its current clean price (i.e. excluding accrued interest).
However, current yield is only a rough and ready measure of a bond’s return. As the bond approaches maturity, its price will converge towards par, and this will affect the yield. The effect is called pull to par or reduction of maturity. The size of this effect is given by the difference between the yield to maturity and the current yield, and may be regarded as the return due to capital gains or losses between the time the bound is bought and the time it matures. Therefore, a more accurate way to measure non-term structure return is to express it in terms of coupon return (running yield), plus return due to the passage of time (pull-to-par return). The pull-to-par return is given by the difference between the yield-to-maturity return and the current yield return:
FIA offers several options for calculation of yield returns:
- NONE: Suitable for portfolios of zero-coupon or equity-type securities, where there is no yield return;
- AGGREGATED: All returns from accrued interest are assigned to a single category, which measures the return from yield to maturity;
- PULL_TO_PAR: Returns from accrued interest are split into current yield and pull to par yield, with the sum of the two returns equal to the yield to maturity return.
Use the CouponDecomposition flag in the configuration file to indicate which option is required. If this flag is not set, FIA uses the AGGREGATED setting.
The user has the option of supplying a yield to maturity for each record in the weights and returns file. If a yield to maturity is not provided, it is calculated automatically using a numerical iterative routine.
Consider a bond that has a single cash flow one year in the future. The yield curve is steeply sloped at the 1-year maturity point, but flattens out at longer maturities.
Suppose that market conditions do not change for a month. At the end of this month, this one-year bond will have become an 11 month bond, and the yield used to price this security will now be read from the 11 month point rather than the 12 month point on the yield curve. Since the yield curve is downwards sloping, the 11 month yield will be lower. Since the yield is lower, the price will be higher, and a positive return will have been generated.
This strategy is sometimes called riding the yield curve, as it is most effective when a security’s cash flows are positioned at maturities where the curve is most steeply sloped.
Note that this return has not been generated by movements in the market, since we explicitly assumed that market conditions were unchanged. Nor has it been generated by elapsed time, because the return is generated entirely by a change in yield. Roll-down is therefore distinct from either source of return, and should be measured separately.
Use the RollDownAttribution flag in the configuration file to indicate whether roll down return should be shown on attribution reports. If not shown, roll down return is added to the residual return for each security.
Term structure attribution
First principles versus perturbational attribution
FIA offers two approaches to attribution of returns generated by curve movements.
- First-principles pricing: by ‘first-principles’, we mean valuing each cash flow generated by the security at the appropriate maturity on the yield curve, and summing the total value of all these discounted cash flows to produce a market price. The change in price caused by pricing with two different curves provides the return due to risk. This approach requires a zero coupon curve for each security, and sufficient information about each security to calculate its cash flows, such as maturity date, coupon and coupon frequency.
- Perturbational attribution, using perturbational returns and risk numbers. This approach uses risk numbers, such as modified duration and convexity, as a proxy for a pricing model.
To run this type of attribution, perform a Taylor expansion on the price of a security and remove higher order terms, which gives
a Taylor expansion on the price of a security and remove higher-order terms, which gives
Writing the return of the security as
this leads to the perturbation equation
where the last term denotes higher-order corrections that may be ignored, and
The terms and measure first- and second-order interest rate sensitivity. These are conventionally referred to as the modified duration and convexity of the security, and are often called risk numbers.
The modified duration of a security measures its price sensitivity to parallel changes in the level of the yield curve; more specifically,
where is return, is modified duration, and is the change in yield of the security.
Perturbational attribution implicitly assumes that the risks of a security can be modelled by representing its cash flows as a single bullet payment, rather than as a stream of individual cash flows over time. For some securities such as coupon-paying bonds, this gives highly accurate results, since the bulk of the security’s cash flows are concentrated at maturity. For other securities such as mortgage-backed securities, where cash flows are more widely spread over the security’s lifetime, the approach is less ideal, although lack of information about a complex securitised security may compel its use.
At what point on the yield curve should we measure changes in the yield curve for perturbational attribution? The maturity is not suitable, since this assumes that changes in price are only affected by the cash flow at maturity. Our preferred measure is the security’s Macauley duration, which is related to the modified duration by:
where MD is modified duration, D is Macauley duration, y is yield to maturity, and n is the coupon frequency for the security. Macauley duration is a cash-weighted average of the term to maturity of a bond, and provides a suitable average at which yield changes should be measured. This measure is also suitable for instruments such as floating rate notes, which may have a maturity many years in the future but a very short modified duration due to frequent coupon resets.
General approach to yield curve attribution
FIA employs a successive yield curve method for both approaches. To run attribution, the program uses a number of yield curves, each representing the effect of a progressive change in the curve when different sources of risk are added in. The effect of changes in each yield curve is then computed on the security’s return For instance, suppose we are modelling changes in the sovereign curve in terms of shift, twist and other types of curvature.
- The first curve is the level of the curve at the start of the calculation interval.
- The second curve is this initial curve plus the parallel shift that occurs over the calculation interval.
- The third curve is the initial curve, plus the parallel shift, plus the twist shift, over the calculation interval.
- The final curve is the curve at the end of the interval, which contains shift, twist and other effects.
The time at which each price is calculated is at the end of the interval. All time effects are calculated in the previous section, so we explicitly exclude time effects from this part of the calculation, which is specifically designed only to measure returns due to changes in the yield curve.
The sum of these changes (time and yield curve) is the overall change in the yield curve over the interval. Any discrepancy seen between the return calculated from these changing prices and the actual return will be due to credit, market noise, or other effects. Further changes due to movements in the credit curve, MBS repayment rates and other risk effects can be added in a similar manner.
Calculating parallel shift
Parallel yield curve shift is regarded as one of the major drivers of fixed income fund performance. There are good reasons for this. Principal component analysis (eg Phoa, 1999) shows that parallel curve shifts usually account for at least 90% of the return of managed bond funds from sovereign yield curve effects.
This is reflected in the widespread use of modified duration as a proxy for a security’s risk exposure to curve movements, where modified duration represents the sensitivity of the security’s return to parallel curve movements. However, there is no standardised, accepted way of calculating parallel shift.
FIA offers two ways to calculate the average level of the yield curve, and hence changes in its average level:
- Arithmetic average: A simple average of all yields is calculated. This is simple and widely used, but tends to amplify the effects of changes at the short end of the curve if there are more points supplied at the short end – which is often the case.
- Trapeziodal integration: Calculates the area under the yield curve, and divides by the difference between the largest and the smallest times. This is probably the most accurate way of measuring parallel shifts, as it removes sensitivity to variable sample spacing along the term structure.
Other types of averaging may be introduced in future.
Sovereign curve return
Sovereign curve return is generated by changes in the AAA sovereign curve (reference zero-coupon yield curve). FIA offers the following options:
No effect calculated
Any returns due to changes in the sovereign curve is assigned to residual. This approach is appropriate for portfolios that are purely credit driven and have no exposure to changes in the sovereign curve.
Aggregate effect calculated
Using the yield curve at the start and end of each calculation interval, each security is repriced using both curves and the return is generated. No sub-effects are calculated and a single return figure is generated. This type of decomposition is suitable for the simplest possible attribution.
Calculates return due to (i) parallel shifts in yield curve, (ii) non-parallel shifts
For simple duration attribution, FIA calculates three yield curves:
- The curve at the start of the interval
- The starting curve, plus the parallel change in the curve;
- The starting curve, plus parallel changes, plus non-parallel changes (equivalent to curve at end of interval)
Each security is priced on each curve to generate three prices . Return due to parallel shift is then given by , and the return due to non-parallel shift by . The sum of the two terms is , which is the overall return of the security.
Shift, twist, curvature return
FIA calculates four yield curves:
- The curve at the start of the interval
- The starting curve, plus the parallel change in the curve;
- The starting curve, plus parallel changes, plus twist changes
- The starting curve, plus parallel changes, plus twist changes, plus other higher-order changes (equivalent to curve at end of interval)
Each security is priced on each curve to generate four prices . Return due to parallel shift then calculated by , return due to twist shift by , return due to higher-order shifts by . The sum of the three terms is , which is the overall return of the security.
Key rate duration return
A key rate duration analysis isolates the effects of changes at particular maturities along the yield curve, rather than measuring the effect of different types of movements.
Key rate duration analysis may be appropriate when running attribution on portfolios of securities that have cash flows spread across a range of maturities, rather than having the bulk of their yield curve exposure concentrated at maturity. Securities in the former category include mortgage-backed bonds and other amortizing securities, and related securitized securities.
In order to run a KRD analysis on a given security, FIA uses a zero coupon yield curve at the start and end of an interval, and a set of reference maturities.
- The security is first priced off the start curve.
- The start curve is modified so that its level at the first reference maturity is changed to the corresponding level at the end curve. Yields that lie at or beyond neighbouring reference maturities are left unchanged, while yields that lie in the interval adjoining the current reference maturity are linearly scaled. The security is then priced off this intermediate curve.
- The start curve is then successively modified so that its value at the nth reference maturity is changed to the value from the end curve, as described above. At each change, the security is priced using the new curve.
- At the end of the process, the pricing curve is identical to the end curve.
The return due to the changes in the prices is now calculated. The sum of the returns will equal the overall return for the security over the interval, and the individual sub-returns are generated by changes at the given reference maturities.
The sensitivity of a security's price to changes at a particular maturity is measured by the key rate duration, just as the sensitivity to parallel curve shifts of a security's overall price is measured by the modified duration. FIA does not currently export key rate durations, but this feature may be introduced in future releases.
CCB attribution uses the Colin-Cubilie-Bardoux algorithm to calculate the twist and curvature movements of a yield curve. This algorithm uses a conventional approach to calculating the parallel shift of a yield curve, but performs a least-squares fit of a first-order polynomial to calculate the twist of the curve. This removes many of the inherent problems involved when fixed twist points are defined.
Principal component analysis (PCA) uses a suitably large number of historical yield curve changes to determin a small set of basis functions that can be linearly combined to represent these curve movements in the most economical way.
This is accomplished by forming the variance-covariance matrix V from the sample of spot rate changes at the N maturities selected. If we then calculate the N orthogonal eigenvectors of V and rank by order of eigenvalue size, the highest ranked eigenvector forms a basis function that explains as much as possible of the observed curve motion in terms of a single vector. By using a combination of this vector and lower ranked eigenvectors, the underlying data can be approximated to any degree of accuracy required.
The variances of the principal components are given by the magnitudes of the eigenvalues, so that the eigenvector with the highest value has the most explanatory power on the underlying data. If the values of the majority of eigenvalues are low, then this indicates that the underlying data can be closely modeled by a small number of functions, which represent some underlying structure in the data. PCA is therefore a useful technique for reducing the dimensionality of a modeling problem. In particular, PCA has been found to work well on yield curve changes (Phoa, 19981; Barber, Copper, 19962), since in practice practically all yield curve changes can be closely approximated using linear combinations of the first three eigenfunctions from a PCA.
PCA on historical yield curve data shows that curve movements fall into a number of fairly clearly defined types. Typically, the first eigenfunction is close to a flat line, the second rises monotonically (but is seldom a straight line), and the third imposes some curvature motion. These functions are usually interpreted as shift, twist, and curvature.
However, these movements are typically slightly different from more conventional interpretations of these terms. The shift movement from a PCA is usually close, but not identical to, a parallel curve shift, and the twist movement is not uniform across all maturities. For these reasons, a PCA may not directly represent investment outcomes in terms of the decisions that were taken by the trader.
Credit and sector curve attribution
FIA allows multiple curves to be assigned to each security.
Consider a corporate bond that has a AA credit rating. The bond's cash flows are priced off the AA zero curve instead of the AAA curve, and the bond's price is therefore dependent on both the level of the sovereign AAA curve and the spread between the AAA curve and the AA curve.
In general, FIA measures the return contribution made by changes in spreads between the curves assigned to each security. The process is similar to that for duration or STB sovereign curve attribution, but measures the extra return generated by changes in the set of sector or credit curves.
Residual return is the difference between the sum of all calculated returns and the actual return, as supplied in the portfolio file.
Ideally, residual returns will always be zero. In practice this is unlikely to be the case, as security-specific factors may lead to slight differences between the calculated and the actual return. If a residual return is significant, this may be indicative of a pricing issue.
FIA can assign residual return to any category required.
In some cases, a new security type may need to be included in a portfolio even though it is not a type that is included in the current security library.
FIA allows this case to be handled by use of the unattributed security type. If a security type is described as unattributed, it still appears in the attribution analysis but no attribution is performed. Instead, all of its performance contribution is written to the Unattributed return category.
This can be useful when, for instance, a new security type has been purchased that has not yet been classified, but its overall performance contribution is relatively low, and therefore will not materially affect the conclusions of the attribution report. Depending on your requirements, you may find this approach preferable to halting the production of attribution reports until the security has been classified.
Paydown return only applies to securities of class SINKER.
Paydown return is generated by amortizing securities such as MBS and ABS, where the principal of the bond can be returned (or paid down) faster than expected under a normal amortization schedule. This paydown is generated by the underlying assets being paid back ahead of schedule, which may be due to homeowners refinancing their mortgages, making extra payments to decrease the life of their mortgage, or other effects.
The sign of the paydown return depends on whether the MBS is trading at a premium or a discount. If the security is at a premium, the paydown return will be negative, since the principal paid will be worth less as cash than if it remained invested in the security. If the security is at a discount, the paydown return will be positive, since the cash will be worth more in the hand than if it were invested in the discounted security.
Paydown return is a separate source of return, distinct from carry return, market return and credit return. It should therefore be placed in its own category on attribution reports.
Paydown return is given by
where is the change in the holding of the security over the payment period, and p is the security's market price. is provided in the bond factor field for the security, and this can be varied over time using FIA's effective date functionality.
Cash deposit return
Cash return is only generated by securities that have OpenRisk function Interest specified in the Risks field.
It is calculated as the product of the annualised interest rate read from the associated yield curve, times the length of the current interval, expressed as a fraction of a year:
Cash return is only generated by securities that have OpenRisk function Inflation specified in the Risks field. It is calculated as the product of the annualised inflation rate IR supplied in the index file (typically updated quarterly), times the length of the current interval, expressed as a fraction of a year:
Other types of return can be specified by the user using the OpenRisk interface. Such returns include those due to optionality.
1 Phoa, Wesley, Advanced Fixed Income Analytics, Frank J. Fabozzi Associates, 1998
2 Barber, Joel R., Copper, Mark L., Immunization Using Principal Component Analysis, Journal of Portfolio Management, Fall 1996