Uh oh. There’s that word again — convexity.
Here it is in the Wall Street Journal:
[FTN Financial’s Walt] Schmidt said mortgages are performing well, considering 10-year Treasury notes are yielding close to 3.2%. He said that at some point convexity hedging will become a factor, as he expects refinancing levels to pick up if mortgage rates continue to hover below 5%. However, it wasn’t in evidence today, Schmidt said.
And here it is at bond-blog Across the Curve:
I have had some conversations regarding a convexity avalanche and the consensus is that lower will rates will foster buying but participants with whom I spoke thought that there would not be enough prepayments to spawn a huge refi wave.
So what is all this convexity stuff?
Readers of this blog will remember last June — when we spoke of something called “maximum negative convexity“. As a reminder, mortgage-backed securities are said to have negative convexity since they tend not to rise in price as much as a normal bond as interest rates decrease.
When interest rates are very low, as they are now in the US, homeowners often take advantage of the rates by refinancing and paying down their old mortgages. When that happens MBS owners find their bonds are repaid faster than expected. To offset that, the MBS investors usually buy longer-dated assets such as US Treasuries. When interest rates start to rise, the reverse occurs and MBS investors start selling longer-dated stuff.
All this is called convexity hedging/buying/selling — and it was the latter which was a concern back in June. MBS rates were nearing a level that might have necessitated massive convexity selling by investors, prompting fears of a significant sell-off in Treasuries (something like $149bn worth of 10-years, according to some estimates).
This time around, however, bond yields have been falling significantly (prices have been rallying) but convexity hedgers have yet to appear. Why not?
Here are some thoughts from the fixed income team at Deutsche Bank:
One factor that has been conspicuously absent has been convexity-related buying in any meaningful size. The crucial questions are why this has been the case, and whether further rallies in yields would cause convexity buying to show up and take interest rate levels down another notch. In our view, sizeable convexity buying will likely not emerge, other than in marginal size as a result of minor rebalancing by some convexity players.
Deutsche sees four reasons:
First, not only does the Federal Reserve now own 23% of the outstanding 30Y agency MBS universe, but they also own the coupons (4.0 and 4.5%) that have the maximum negative convexity.
Second, the quality of price discovery in the agency MBS market has significantly deteriorated due to the involvement of the Federal Reserve. The total outstanding amount of 30Y agency mortgages has increased by $470 bn, or 13.5%, since the beginning of this year. At the same time, the MBS holdings of the GSEs and commercial banks have remained almost unchanged. The implication is that the Fed’s portfolio growth has probably come from real money managers or other active participants (as opposed to put-away buyers). The float in the MBS market is thus significantly smaller than it used to be. Without clear price discovery it has been difficult for banks and GSEs to determine their risk.
Third, the estimates of duration risk and convexity risk are heavily dependent on government policies, rather than interest rates. The institutions that had typically hedged their MBS-related convexity exposures have been the GSEs and mortgage servicers. Over the last two years, it has become increasingly difficult to measure the duration, convexity, and volatility risk of MBS portfolios, due to the influence of government policies. For example, in the absence of the HARP (the Obama Administration’s Home Affordable Refinance Program), mortgages should be quite long in duration as 32% of agency MBS would be on homes that are 80 to 125 LTV. If this program were to be fully implemented, prepayments could start to skyrocket. So far, as observed during the last brief refi window in April, we have seen very little prepayments go through as a result of the Administration’s refi plan. There has been a tremendous amount of uncertainty regarding how much the HARP will affect the coming refi window. Thus institutions that desire to hedge their servicing or MBS portfolio will find it very difficult to decide on the risk parameters such as durations, convexity, etc., and could thus essentially refrain from hedging instead.
Fourth and finally, the change in duration has not affected the institutions that hedge their mortgage portfolios because the recent rally was accompanied by a large decline in short-dated implied volatilities (gamma) but relatively small changes in long-dated implied volatilities (vega). Thus the duration change of high-coupon mortgages was offset at least in part by the reduction in the level of gamma volatility. On the other hand, in the case of 4s and 4.5s the gamma decline had a lesser offsetting effect on the duration. Also, the MBS spread tightening was felt more in these latter low-coupon mortgages since their spread convexity is higher. But since these low-coupon mortgages are owned mostly by the Fed, which has no need to hedge, the convexity impact has been mitigated. The recent bull flattening in the Treasury curve has had the same effect, with the lower coupon mortgages similarly having the maximum change in duration, but little broader convexity impact.
So will convexity hedging eventually appear if MBS rates/bond yields continue to fall?
DB thinks not — the structural changes in the MBS market have simply been too great:
Going forward, we don’t think convexity will reappear as a major driver of rates even if we break out of this range in the lower rate direction. Further declines of gamma in the rally scenario will cause a disproportionate duration impact on the Fed’s portfolio, rather than on the bank and GSE portfolios that are likely to be convexity hedged. In addition, prepayments due to buyouts, which are now 5- 10% of the total CPR for the higher coupons, will also be less interest-rate dependent — on the contrary, these kind of prepayments adds positive convexity to the mortgage portfolio. Finally, the factors that have moderated the impact of the Obama Administration’s refi plan, such as the reluctance of banks to initiate prepayments, are still continuing to do so.
One to watch, in any case.
And we are still left to wonder if the Fed — “having no need to convexity hedge” as DB puts it — will eventually exceed the limits of its own duration and extension risk.
Related links:
Negative convexity at the Fed - FT Alphaville
The Fed’s asset-liability mismatch - FT Alphaville
