Earlier, we noted that with spec Treasury net shorts hitting an all time high in both the 2Y and ultra-long futures...
... just as the long-running correlation between equities and yields finally snapped this week as higher yields are now seen as a risk to stocks..
... traders are increasingly worried, after the worst week for stocks since Jan 2016, what further Treasury selling could mean for equities as a result of a potential violent deleveraging of risk-parity strategies, which will be forced to significantly reduce their gross exposure following last week's drubbing.
Yet with everyone's attention now on deleveraging risk-parity funds, as well as momentum-chasers and CTAs "wreaking havoc" in the short-end, there is one potential source of Treasury selling that the market has largely forgotten about: convexity hedgers.
Luckily, thanks to Barclays' Dennis Lee, here is a reminder of the selling overhang potential from convexity risk.
As Barclays writes, with the 10y having jumped 43bp since the start of the year, investors have begun to question how much convexity-related selling could result from the extension in mortgages. Since the start of the year, the Bloomberg-Barclays MBS index has extended 0.71 years
While this is sizeable, it is still far less than the 1.7yr extension in the index that occurred between November 8, 2016 and December 15, 2016, in the weeks following Trump’s election victory. Then again, the move may only be starting.
So what is the potential selling overhang as a result of recent rate moves?
Based on Barclays' model, at the start of the year, a 50bp sell-off in rates would have led to a $667bn extension for the MBS universe expressed in 10y Treasury equivalents (chart below).
This compares with $849bn of 10y Treasury equivalent extension on November 8, 2016, the night of the presidential election, for a 50bp sell-off in rates. In comparison, we the British bank estimates that a similar sell-off in rates today would extend the MBS universe by only $408bn 10y Treasury equivalents, suggesting that further extension risk in MBS is fairly modest.
Alleviating the risk of a selloff is the fact that a smaller portion of the MBS universe dynamically hedges its duration exposure than it did before 2008. In particular, the Federal Reserve ($1.7trn in MBS holdings), many money managers (~$700bn in MBS holdings), and several overseas investors (~$800bn in MBS holdings) do not actively hedge duration in their MBS portfolios.
Furthermore, Barclays notes that empirical evidence also suggests that the MBS universe is exposed to less extension risk today than it was in late 2016. Part of the reason is that the average mortgage rate for the universe has continued to decline over the past several years.
For example, the average WAC among post-HARP FNCL pools stands at 4.26%, down from 4.33% at the time of the November 2016 elections (Figure 5). With the prevailing 30y no-point Freddie Survey Rate already at 4.33% as of this week, the majority of the universe now has little or no refinancing incentive.
This is a far cry from what occurred in November 2016. During that time, the 30y no-point Freddie Survey rate was approximately 3.64%, suggesting that a significant portion of the MBS universe had refinancing incentive. Indeed, we estimate that close to 50% of the post-HARP FNCL universe had at least 50bp of refinancing incentive at the time of the elections (Figure 6). At the start of 2018, this figure was only 15%, and today, we estimate that only 10% of the post-HARP FNCL universe has at least 50bp of refinancing incentive.
As such, Barclays concludes, at this point, even a further substantial sell-off in rates would not materially reduce the percentage of refinanceable borrowers.
In summary, this is good news for those worried about further self-reinforcing selling at least out of convexity hedgers. The bad - or at least so far undetermined - news, is that forced selling by "everyone else" may be sufficient to tip the overall market, mostly via vol-sellers, CTAs and risk-parity funds, into the next correction.