Resources

Build Better Curves Using Today's Market Data -
Q & A session

Q: Build Better Curves Using Today's Market Data

A: The basis spread adjustment workbook enhances FINCAD curve building functionality, and is based on FINCAD functions. FINCAD has made this enhancement available to all customers using FINCAD Analytics Suite 2009 for Excel or FINCAD XL version 10 or higher with a current support subscription.

Q: How do we get a copy of the basis spread adjustment workbook?

A: Contact your FINCAD representative, by phone or email at info@fincad.com.

Q: What kind of documentation does the workbook come with?

A: This workbook, like all FINCAD products, is extensively documented. Documentation provided with the workbook includes an explanation of the rationale for creating this workbook, instruments affected, and step-by-step instructions on how to set up the workbook.

Q: Is the basis spread adjustment available in FINCAD Analytics Suite for Developers?

A: Not yet, but it will be. The flexibility of FINCAD Analytics Suite allowed us to provide a timely response to this new market situation. While this solution has not yet been built into our software development kit, software developers can use the workbook as a prototype to develop solutions with FINCAD Analytics Suite for Developers (with the assistance of FINCAD Client Services if required). Also note that FINCAD is planning to further enhance our basis spread adjustment capabilities in our next software releases for both our Excel-based products and our software development kits - which are currently scheduled for the second half of this year.

Q: What types of instruments are affected by 'non-standard' tenors?

A: Any instruments where the tenor of the reference rate differs from the reference rate tenor of the standard par swap rate quotes used to build a swap curve. A quick list includes: interest rate swaps, floating rate notes, swaptions, and exotic IR instruments.

Q: Are there differences between European, US and Asian curve methods?

A: The curve building methodology is the same. The differences are the data used and the market conventions.

Q: What should I use if futures data is not available?

A: If futures data is not available, try to find forward rate agreement data. These are usually available for most currencies, but should only be used if the instruments are sufficiently liquid.

Q: If I have existing workbooks can I use this basis spread adjustment workbook with them?

A: Yes, there are two ways to use this workbook in conjunction with existing workbooks. One way is to build the basis spread adjusted curve and copy and paste the values into the existing workbook, having the necessary function input reference the curve. Another way is to insert the curve template into existing workbooks. Please contact your FINCAD representative for the article, "Mastering Master Curves" to get detailed instructions.

Q: When constructing the yield curves as combination of the three (cash, futures or forward, swap) curves, how is the time horizon for the middle part of the curve determined? What are the reasons for different horizons for different currencies?

A: There is not a single correct set of data that should be used for curve building for all applications. The data used will depend on the rate quotes available in a given market, and the liquidity of the underlying instruments. As an example, a common set of data for constructing a USD swap curve is to use cash deposit rates out to 2m or 3m, Eurodollar futures out 2 to 4 years, and then par swap rates.

Q: Can I download the PowerPoint so I can I ask questions later?

A: The presentation from the Taming the Yield Curve webinar is available on our website at: www.fincad.com/derivatives-resources/webinars/taming-yield-curve.aspx.

Any additional questions may be sent to: info@fincad.com.

Q: What about serial futures contracts that overlap?

A: Any overlap in the futures contracts will be dealt with by interpolating the required discount factor. As an example, consider the Jun09 contract and the Jul09 serial contract. The Jun09 contract is effective from 17-Jun-2009 to 17-Sep-2009, and will give the discount factor for these dates. The Jul09 serial contract is effective from 15-Jul-09 to 15-Oct-09, so when bootstrapping, a discount factor will be required for 15-Jul-09, and the futures rate will then determine the discount factor for 15-Oct-09. The discount factor for 15-Jul-09 is obtained by interpolating between the 17-Jun-09 and 17-Sep-09 discount factors that are already calculated in the bootstrapping process.

Q: Can using different data sources for different sections of the curve lead to different credit quality in different parts of the curve (e.g. due to collateral quality or the counterparty)?

A: Yes, it would be a problem to use rates in different parts of the curve that reflect differing levels of credit risk. We recommend using exchange-traded futures, and rates for par swaps with low counterparty risk (or managed risk using collateral agreements).

Q: Is the basis spread curve available in FINCAD.net?

A: The basis spread curve is not currently available in FINCAD.net.

Q: Are instructions available to pull futures information automatically from Bloomberg? Is there a version of the new workbook that includes links to Bloomberg?

A: FINCAD workbooks can currently be linked to BLP futures data. Ask your FINCAD representative for the FINCAD article, "Mastering Master Curves".

Q: After constructing the basis spread curve, how can we use this curve in other worksheets, where the discount and amortizing curve tab is already set up?

A: Please ask your FINCAD representative about incorporating the basis spread curve into other workbooks. They can determine whether we have the one you're looking for, if there is one being developed, or if it would require custom work.

Q: Which interpolation would you generally recommend?

A: We recommend the "constant forward rates" bootstrapping method when building the curve, and "exponential" interpolation when using the curve. This combination is a reliable workhorse in most situations, that works well for analyzing vanilla interest rate derivatives (e.g., swaps), both in terms of pricing and key rate analysis. If a smoothed forward curve is required in order to achieve a good calibration of an interest rate model, then use the "quadratic forward rates" bootstrapping method when building the curve, and "cubic spline" interpolation when using the curve. There is a FINCAD Newsletter article on this topic at www.fincad.com/derivatives-resources/articles/curve-building.aspx.

Q: What kind of volatility convexity adjustment parameter would you recommend for the futures?

A: We provide functionality for calculating a convexity adjustment based on the Hull-White or Ho-Lee one-factor term structure models. The FINCAD function to use for this purpose is aaEDFut_CnvxAdj_HW.

Q: Under FASB 157, the value should also consider illiquidity spreads and counterparty risk adjustments. Should these adjustments be built into the curve or separately?

A: The base curve should be constructed first and then any liquidity spreads or adjustments should be applied to the base curve.

Q: Is it ever appropriate in any cases to not factor in tenor basis spreads in curve construction when valuing a derivative that is say pay USD fixed vs. receive USD 6M. I.e. do any banks not factor in the tenor basis when pricing or valuing a trade?

A: A basis spread adjustment needs to be made to any curve that will be used to imply forward rates that are of a different tenor than the reference rate of the instruments used to build the original curve. For example, if pricing a swap that references 6M USD LIBOR, you will need to adjust the curve used to imply forward rates because the swaps used to build the standard USD LIBOR curve are referencing 3M USD LIBOR. We cannot speak to what all banks are doing, but FINCAD is trying to level the playing field for all sell-side and buy-side users.

Q: Where is the market going as far as using Futures vs. FRAs?

A: In many jurisdictions, one is more liquid than the other. Use the most liquid instruments.

Q: When valuating derivatives as at 31 March 09, there are no evident peaks in the forward curve. Does that mean we are safe with using deposit and swap rates only?

A: Using just deposit rates for the short end of the curve and then swap rates for the long end can result in a forward rate curve that does not have any obvious inconsistencies in the short end. In this case however, you would want to use only deposit rates with maturities that are less than or equal to the tenor of the swap reference rate. It is advisable to include futures in the curve building process, but it is not mandatory and there is not a single correct set of data to use for all applications.

Q: Can you show in the workbook how I can specify discount using 3M LIBOR curve but forward using 6M or 1M LIBOR curve?

A: FINCAD's pricing/cash flow functions usually have two inputs:

  1. Discounting curve – for discounting cash flows
  2. Accruing curve – for calculating implied forward rates.
Referencing the output of the basis spread workbook to 2) and the standard curve to 1) will ensure the correct curve for the desired purpose.

Q: On the curve construction workbook, do I have a place to specify which rates to use? For example, where can I specify the future starts? Also do I need to input my own convexity adjustment or do I input all the points and FINCAD will take care of the interpolation itself?

A: In the curve template, there is a "use point" drop down that allows the user to specify whether to use that point or not. This switch allows the user to turn on and off points used in constructing the curve. The swap curve building function itself will not apply any convexity adjustments to input futures rates, however in our curve workbooks this is done for the user in the spreadsheet if they choose to use a convexity adjustment in the "Futures Data Settings".

Q: How do you combine different instruments (e.g. cash, futures, swaps) to build a single yield curve?

A: Our curve building function has the capability to extract information from market quotes. It is relatively straight-forward for cash and futures. Bootstrapping is required to extract information from par swap rates. Please see one of our previous Newsletter articles, "The Art and Science of Curve Building": www.fincad.com/news-events/assets/pdfs/dec05/curve_building.pdf.

Q: Is the reason we use the basis curve in your example because the swap used 1M LIBOR and not 3M LIBOR which is the standard tenor that the swap curve is built against?

A: Yes. In the example shown, the instrument being priced references 1M LIBOR. A basis spread adjustment needs to be applied to the curve in order to imply forward rates for 1M LIBOR.

Q: Where do you get the basis spread information?

A: You can get the basis spread information from your market data provider. FINCAD will be able to provide this information through FINCAD Market Data by the end of June. You could also get the information from Bloomberg screen WCV.

Q: Any insights into why the futures based curve diverges from the basic curve? Any insight into why there is a basis spread between 1M and 3M LIBOR?

A: Assuming this question refers to the plot on slide 11 of the webinar, the short end of the "no futures" curve uses deposit rates out to 12M, which introduces a basis spread into the short end of the curve as the reference rate of the swaps is 3M USD LIBOR. This results in implied forward rates at the short end being too high, as illustrated in the plot. Single currency basis spreads between rates of different tenors represent a liquidity and credit risk differences between the two rates.

Q: Is the consistency between DF and forward rates preserved when basis spreads are included in the curve construction?

A: The Basis Spread Curve created by the workbook encodes information about forward rates, and as such is intended to be used as the "accruing curve" input to FINCAD functions that calculate price and implied cash flows. The Basis Spread Curve is not intended to be used for discounting.

Q: Does the basis spread workbook also work with Prime Spreads, CP Spreads, and other curves that trade to a spread to LIBOR, not just LIBOR basis swaps?

A: No, the workbook is set up to produce a curve that will price the set of given basis swaps to par when used to imply forward rates for the leg that references a different tenor than referenced by the par swap quotes.

Q: In the case of overlapping rates, for example 3M LIBOR cash rate and the first Eurodollar futures contract, can the functions interpolate rates in order to construct a clean discount factor curve?

A: The functions can interpolate linearly from discount factors, exponentially from discount factors, or linearly from spot rates.

Q: I noticed in your Excel workbook that you have overnight and one week LIBORs used, and then you go straight to futures for a few years. Swap rates started at 3 years. Is this something that you or FINCAD recommend using? I'm wondering this since this appears that 1- and 3- month LIBORs are illiquid.

A: A common set of data for constructing a USD swap curve is to use cash deposit rates out to the 2m or 3m, then use Eurodollar futures out 2 to 4 years, and then use par swap rates.

Q: After watching your webinar this morning, I have tried to use Eurodollar futures to build my interest rate curve. However, the convexity adjustment for futures requires Hull-White parameters, and the calibration of H-W parameters requires a discount curve. How do you recommend eliminating this circular logic?

A: There are a few suggestions for how to deal with this. The first option is to use calibrated parameters from the previous day for the current day's convexity adjustments. Alternatively, the discount factor curve from the previous day could be used with the current swaption or cap/floor data to obtain the calibrated parameters for the current day. Another option would be to build a discount factor curve using the current day's data without any convexity adjustments, and use this curve for the calibration of the H-W parameters. The calibrated parameters can then be used to calculate the convexity adjustment for futures, which can then be used to properly build a curve.

Q: How many months of LIBOR rates do you recommend using to build the discount curve? 3 months and less?

A: For the short end of the curve, we do not recommend using LIBOR fixings as they do not represent an actual traded rate. We recommend building the short end of the curve using cash deposit rates of maturity up to the tenor of the reference rate of the par swap quotes used. For building the USD swap curve, this would be 3 months and less.