As reported here, Wednesday saw a frantic steepening of the Treasuries curve, a fact that has led some to ask whether the Fed might be losing control of mortgage rates — note the chart from BarCap below.
What to make of this, and specifically Wednesday’s sudden moves?
Well it may all be a bit of a chicken and egg scenario. The fact is there is a growing perception in the market that further down the line, a messy quantitative easing-exit strategy might see the Fed forced to raise rates quite fiercely.
The pricing in of those perceptions inevitably opens up the risk of a rush of convexity hedging in the market. In a nutshell, bond traders late to the game of pricing in the prospect of higher interest rates, only help exaggerate the steepening of the curve via the practice of convexity hedging — this being one of the reasons behind Wednesday’s moves.
Just to give the moves some context, the Across the Curve blogger suggests the current shape of the curve is at its steepest since 1992. As he writes:
The 2year/30 year spread is 366 basis points. The record on that is 369 on October 05 1992 at about 1130 AM. i am a very sick man!!!
Barclays further puts the current moves down to the inevitable refinancing wave fuelled by the current low front-end rate environment — in so much as the prospect of higher rates further down the line has led to an increase in mortgage originations or refinancing deals. They compare it to the situation in 2003, when the sudden emergence of a more hawkish Fed created a similar refinancing wave:Over the course of the 2003 refinance wave, gross issuance of mortgages picked up sharply and mortgage pipelines were at their highest ever (Figure 4). Extension risk in the origination pipeline, defined as the increase in pipeline duration for a 100bp backup measured in 10y equivalents, stood at a historically high $175bn in June 2003. As rates started rising in June 2003, on the back of a hawkish FED announcement, pull-through rates spiked, reaching close to 100%. This caused pipeline durations to increase, forcing the originators to sell duration, which caused a further backup. It is estimated that over the course of the sharp six-week extension, the pipeline duration increased by $260bn.
So as the rush of refinancing took place, a wave of convexity hedging was necessary to bring portfolios into line with new shorter duration rates, and the prospect of longer bond duration rates in the future as rates rise. The activity in itself contributes to a steepening of the yield curve.
Of course, back in 2003, the GSEs accounted for nearly $1,200bn in mortgage assets, all of which were hedged actively for duration and convexity. As Barclays explains, GSEs were specifically required via regulation to maintain the asset liability duration mismatch to less than six months. Accordingly:
To meet the duration gap requirements, GSEs pay in swaps during a rate backup and receive in swaps during a rally, causing swap spreads to be directional with rates. Their paying needs exaggerated the backup in the 2003 episode.
This time round, however, the situation is a little different, mainly due to two reasons:
Industry consolidation: In recent months, there has been extraordinary consolidation in the banking industry, implying consolidation in servicing and origination businesses (Figure 9). This should also point to increased convexity hedging given the transfer of assets from less sophisticated to more sophisticated players.
Composition of mortgage issuance: There has been a persistent shift away from nonagency and hybrid products into agency fixed rates (Figure 10). This trend points to increased convexity hedging needs, as fixed rate agency MBS is much more negatively convex.
Simply put, in 2003, GSEs were the main backstop bid for mortgages, because they held mortgages on a hedged basis. This time round, however, the Fed through its MBS purchases has become the ultimate backstop bid. The difference is the Fed is buying mortgages on an un-hedged basis. According to Barclays this should provide for a much more credible backstop.
However there is one risk. If there is any indication that quantitative easing may be about to end, the market could, through its over-dependence on the Fed start seeing an upward drift in rates. Convexity hedgers would then only exaggerate the move.
Nevertheless, all things considered, Barclays says the scenario should not pose as big a problem as it did in 2003:
Despite the increased convexity hedging needs in the current cycle, we believe that the impact on extension may not be as dramatic as in 2003, with Fed purchases being the key difference. As the market gets a hint of an end to quantitative easing, there should be a backup in rates and significant duration shedding from convexity hedgers. But we believe that the Fed would still have enough firepower at that point to prevent a dramatic 2003- style extension. We would expect the Fed to absorb the duration supply coming from the convexity hedgers, leading to a slow and more controlled extension.
We’ll be watching closely.
Related links:
Treasury investors take “sell in May” adage to heart – FT Alphaville
On the not-unlimited appetite for government bonds – FT Alphaville
The Fed’s asset liability mismatch – FT Alphaville
Convexed – FT Alphaville