Hedging MBS Curve Risk with 30-Year Treasuries

# Should one use the 30y bond alongside shorter tenors to hedge a MBS book?


# Should one use the 30y bond alongside shorter tenors to hedge a MBS book?












Say I build a US treasury curve or swap by bootstrapping 2/3/5/7/10/20/30y on the run bonds. Say I have a prepayment model and an OAS model and I can generate key rate dv01s for the book as dBookNPV / dYield(i) where i is in [2, … , 30]. I can then dynamically hedge by shorting the right amount of all the on the runs to flatten the residual vector of above risk measures.

Assuming the partial derivative above is non trivial for the 30y bond, say it represents 20% of the total, my question is as follows:

Would I create unwanted risk in my hedging strategy if I were to stop using the 30y in my bootstrapping thence pricing model as it would obviously show a zero aforementioned risk measure to the 30y and move down such sensitivity into 10’s and mostly 20’s? Do large mortgage desks at IBs used the 30y at all to price and hedge?

## Answer by Sharad (score 1)

https://quant.stackexchange.com/a/75831

- In general, replacing 30-year Treasuries with a mix of 10-years and 20-years does create some residual curve risk since these tenors are not perfectly correlated. This has been especially visible over the last two years with significant volatility in movements of the yield curve. Also, consistent with your remark above, the exposure of MBS lower coupons to the 30-year is non-trivial: one dealer's OAS model shows that currently the 30-year DV01s of 3s and lower coupons represents around 20% of the total duration.

- It is hard to speak for mortgage desks across broker-dealers since their hedging practices are not publicly disclosed but anecdotally (a) The 10-year (and shorter Treasuries) are still the hedging vehicle of choice for passthrough desks and (b) CMO desks with longer PACs, Sequentials, or Zs in position are more likely to hedge with 30-years than passthrough desks.

- The downside of hedging with the 30-year is that is less liquid than the 10-year; in the past, the long bond was subject to technical squeezes more than the 10-year which made hedgers reluctant to use it. Also, rebalancing is less expensive if one focuses on fewer points on the curve. To get a somewhat recent sense of the relative liquidity of different points on the yield curve, take a look at How Liquid is the New 20-Year Treasury Bond?. For a more recent snapshot of one dimension of liquidity, take a look at the TRACE volumes reported by FINRA (Treasury Monthly Aggregate Statistics)

## Answer by Dimitri Vulis (score 0)

https://quant.stackexchange.com/a/75832

In general, when you have a choice of several hedging strategies, the tools to help you compare hedge effectiveness might include:

Compare the VaR (or extended shortfall) of the portfolio being hedged with various strategies and without hedges. It would be helpful if your VaR calculator can output "Component VaR" by market factor type. See that the component VaR from, in your case, interest rates decrease. If the component VaRs from other market factors inctease, e.g., if you considered treasury futures and introduced VaR from cash-futures basis, then that is still within your appetite. If it can't output component VaR/ES, then you're limited to seeing the overall decrease, and should try to get that enhanced.

Look at the time series of the P&Ls of the portfolio and the proposed hedges. It would be helpful if your P&L explain / attribution analysis can separate the P&L from the market factors being hedged from the rest. Look at their correlations. Ponder when the correlations break, and whether things revert afterwards. I hope that as part of your ongoing performance monitoring of pricing models you already run P&L explain, and observe that some models, including prepayment models, can leave much unexplained when markets behave unusually.

I also hope you already have a comprehensive library of market stress scenarios. Ponder what your portfolio and hedges might do under these scenarios. Among others, consider various "flight to quality" scenarios, in which everybody panics, and wants to hold cash treasurys, rather than any futures or asset-backs, as happened, for example, during Lehman bankruptcy or the start of the Covid lockdown.

You should do reverse stress analysis, i.e. look for plausible scenarios that might hurt you. You can look at the historical volatility of the 20Y-30Y spread, and figure how many historical standard deviations move would make a difference in which one you use that you'd care about.

If you were to compare cash treasurys v treasury futures v SOFR futures v IR swaps as possible hedging instruments, then you should compare all the associated costs - the carry, including financing; the cost of each transaction times the number of transactions, the cost of setting up a new instrument if you're not already trading them.

If you choose cash treasurys, I don't see why you want to limit yourself to on the runs. Do you intend to unwind an existing position and get a new one every time a new on the run appears, incurring transaction costs? I've known a few people who used US treasurys as IR hedges, but I don't think they unwound their hedges just because they became off the run.