Probably one of the more interesting aspects about Friday’s rip-roaring 22 handle rally that took the S&P back positive for the week was that the bond market didn’t even budge. The 10YR yield closed at 1.49% not far from the intraday low printed when Wednesday’s auction saw record demand
from direct bidders taking down 45% of the deal. The “when-issued” was trading at 1.52% pre-auction with the results coming in 6bps at 1.459% matching the intraday record low yield hit after the weaker than expected May NFP report on June 1. Unlike indirect bidders (foreign central banks) and primary dealers, direct bidders are real money accounts like insurance companies and pension funds that may include mortgage investors.
Monday the mortgage market was hit with some surprising June prepayment data. From a July 9 Nomura report titled "June Prepayment Commentary":
Sharp Increase in Lower Coupon Prepays
Prepays on cuspy lower coupons increased more than expected, with 2011 FN 3.5s increasing 96% to 18.9% CPR and 2010 FN3.5s increasing 46% to 17.5% CPR from May to June. This sharp increase appears to be driven by a large increase in TPO (third party originators) prepays while the increase in non-TPO prepays has been relatively benign. Despite being less seasoned and having a slightly lower TPO percentage, 2011 vintage 3.5s are currently faster than 2010 vintage 3.5s, likely due to the 20k higher loan size on the 2011 vintage.
Hmmm... Fannie Mae
(FNMA) 3.5s are mostly comprised of 4.0% 30YR mortgages. With recent mortgage yields making new lows at 3.5% this suggest borrowers are either refinancing loans just made over the last couple of years to save 50 basis points, paying down principal early on new loans or there is increased turnover in the housing market.
Thursday I received the following comment from one of my bond dealers and his CMO (collateralized mortgage obligation) trader:
Another very busy day in CMO activity yesterday. We’ve had a rush of buyers looking for structure to protect against prepays and lower portfolio durations. Many have moved out on the curve some in effort to grab more yield. There’s not a lot of extension fear... (emphasis mine)
Translation: Despite interest rates at record lows, MBS (mortgage-backed securities) investors are focused on protecting against falling rates rather than rising rates.
I don’t know much about the specifics of the mortgage market, but I do know interest rate risk when I see it so let me offer a simplified explanation of the risk embedded in MBS securities. Because the borrower has the right to prepay his mortgage at any time, the mortgage holder is effectively short a call option. Because you are short this call, mortgage securities exhibit what is known as negative convexity. For this reason, when you are long an MBS, you are short interest rate volatility. Convexity and its derivative cousin gamma have exponential affects on market price and is a major contributor to extreme market movements.
A bond’s duration is the measure of a move in price for a given move in interest rates. The longer the duration, the more sensitive a bond’s price is to interest rates. In addition, the lower the coupon for a given duration, the more volatile that coupon is for a given move in interest rates. Like gamma, which is the rate of change in the delta or the sensitivity of an option’s implied volatility to moves in the underlying price, convexity is the rate of change in duration or the sensitivity of bond’s price volatility for movements in yield.
As we discussed in the article In the Parallel Universe, Credit Risk Is Interest Rate Risk
, positively convex bonds like Treasuries rise more
in price when yields fall and conversely fall less
in price when yields rise. Negatively convex bonds like mortgages are the opposite and rise less
in price when yields fall and more
in price when yields rise. This is due to the tradeoff in what the CMO trader’s comments imply between a bond that prepays due to falling yields or one that “extends” when yields rise. The average life and duration of a mortgage is constantly changing base on what yields are doing, shortening when yields fall due to faster prepays and extending when yields rise due to slower prepays.
To hedge against this negative convexity, mortgage holders will calibrate this changing duration risk with Treasuries (typically 10YRs). When yields fall, they buy treasuries to add duration and sell when yields rise to reduce duration. You can see that in extreme situations (like record low Treasury yields) mortgage holders can exacerbate movements in the market as they increase demand when prices rise further pushing them higher and vice versa. They can get caught in a feedback loop so to speak. Why is this important?
On Tuesday, Reuters reporter John Kemp
wrote a column titled "The bubble in fear,"
which addresses some of the same issues we wrote about in US Monetary Policy: On a Magic Carpet Ride
in discussing the discrepancy in market pricing of deflation vs. inflation risk.
Buying ultra-low yielding long-dated government bonds that guarantee losses in the event of even moderate inflation or interest rate rises any time before maturity, or leaving cash on deposit at negative real interest rates, as many households and companies across the advanced industrial economies have done in the last three years, is fundamentally irrational behaviour.
The bubble in fear is an obvious reaction to earlier bubbles in more risky assets. In some ways it might be described as an "anti-bubble." But that concept is already used to describe the abrupt collapse in valuations when a bubble bursts or a flash crash. This is much more enduring. It is better described as a true bubble in fear and excess demand for liquidity.
I agree wholeheartedly and have actually been exploring this concept since the Fed launched QE II in 2010. As I told St. Louis Fed President Jim Bullard (cited in Magic Carpet Ride
), my belief is the Fed is actually contributing to this fear by artificially generating negative interest rates, which is not reflecting
fear and slow growth but actually causing
fear and slow growth.
Kemp hints at what I think is very much driving current bond market sentiment and positioning when he referred to George Soros (“The Alchemy of Finance”). In 1985, Soros wrote The Alchemy of Finance
and introduced his theory of reflexivity as it applies to financial markets.
Reflexivity refers to circular relationships between cause and effect. A reflexive relationship is bidirectional with both the cause and the effect affecting one another in a situation that does not render both functions causes and effects.
In Economics reflexivity refers to the self-reinforcing effect of market sentiment, whereby rising prices attract buyers whose actions drive prices higher still until the process becomes unsustainable and the same process operates in reverse leading to a catastrophic collapse in prices. It is an instance of a positive feedback mechanism.
Soros took reflexivity to another level and believed that market price could actually affect the fundamentals that it deemed to discount. My favorite paragraph from the book is something I used in a presentation to a bank’s board of directors in January 2008 to explain what I thought was going to be an ensuing large scale collateral liquidation.
Chapter 3: The Credit and Regulatory Cycle
A strong economy tends to enhance the asset values and income streams that serve to determine creditworthiness. In the early stages of a reflexive process of credit expansion the amount of credit involved is relatively small so that its impact on collateral values is negligible. That is why the expansionary phase is slow to start with and credit remains soundly based at first. But as the amount of debt accumulates, total lending increases in importance and begins to have an appreciable effect on collateral values. The process continues until a point is reached where total credit cannot increase fast enough to continue stimulating the economy. By that time, collateral values have become greatly dependent on the stimulative effect of new lending and, as new lending fails to accelerate, collateral values begin to decline. The erosion of collateral values has a depressing effect on economic activity, which in turn reinforces the erosion of collateral values. Since the collateral has been pretty fully utilized at that point, a decline may precipitate the liquidation of loans, which in turn may make the decline more precipitous. That is the anatomy of a typical boom and bust.
It’s easy to sit here today in hindsight and brush off the analogy to the recent credit bubble as obvious, but at the time this risk was not widely appreciated. Most market participants failed to recognize the role reflexivity played as the credit bubble inflated and thus the role it would play as it collapsed.
Due to convexity, bond markets are especially sensitive to reflexivity and given the US reserve currency status, the US bond market is even more exposed. Consider that as the US finances global trade with excess liquidity and current account deficits, the US creates demand for Treasuries. Ironically the more dollars we export via inflation the more demand we generate for dollar denominated fixed income securities, the most sensitive asset to inflation.
Enter 2010 and QE II when the Fed printed $600 billion to buy $600 billion in Treasuries in an effort to inflate dollar denominated assets and generate negative interest rates. This put reflexivity into hyperdrive igniting the bubble of fear.
By generating excess liquidity and at the same time taking Treasury supply out of the market the Fed initiated a mad scramble for dollar denominated fixed income securities. The more demand they created, the lower interest rates fell which accelerated mortgage prepayments which forced more negative convexity hedging which created more demand for Treasuries which pushed yields lower.
The more demand created for securities in the financial economy, the less supply of liquidity for the real economy which is one reason velocity continued to collapse. Excess liquidity is sitting idle on bank balance sheets in securities ($2.6 trillion) instead of being extended into the economy. Just as Soros posited, the low interest rates are affecting the economic activity they purport to discount. They are driving slow growth.
The other side of a demand for liquidity and deleveraging is a decline in consumption which reduces excess dollars being exported. In Friday’s Economics Bloomberg Brief
economist David Powell reported on the decline in global reserve accumulation.
Five of the 10 largest holders of foreign exchange reserves in the world have reported declines in the level of those assets for the second quarter.
Those figures suggest a sharp deceleration in reserve accumulation on a global basis. The 12-month sum of newly accumulated reserves, adjusted for valuation effects, for the 56 largest holders of foreign reserves for June equals $666.9 billion, according to Bloomberg Brief calculations, with nine of the 10 largest reserve accumulating countries having already released data for the month. That figure compares with a peak of $1667.4 billion in May 2010.
The author also shows the positive correlation between foreign reserve accumulation and the value of the EURUSD. As foreign reserves decline presumably as fewer excess dollars are exported the dollar strengthens. With it foreign demand for dollar denominated fixed income securities. This is similar to the point we were making last week in The US Dollar’s Triffin Dilemma Adjustment Disorder
Just as in 2008, when the economy is being dictated by reflexivity, you are likely nearing the end of the cycle because as Soros says, there comes a point when the stimulative affect (or anti-stimulative affect in the bond market) will run its course. The question for investors is whether the reflexivity in the MBS market is the tail wagging the US Treasury market dog. If the record direct bidding demand at the 10YR auction at record low yields is any indication we could be witnessing the final stages of the reflexive process in the bond market.
This mortgage lending cycle has been a perpetual exercise in refinancing (prepaying) as the three decade interest rate bull market has continuously produced an opportunity for borrowers to lower their mortgage rate. However, conceivably mortgage rates will get to a point where there will be no incentive to ever prepay again because the money is free (we are close), you can’t lower rates to a level where it is cost beneficial to refinance and rates will never be lower in the future. At that point, the current coupon MBS will see prepay risk virtually eliminated (never 100% eliminated) and the average life and duration could rise considerably.
The 2011 FN 30YR 3.5s generally have a duration of between three and five years depending on the composition of the loan pool. If these 4% mortgages are prepaying, it wouldn’t seem to take much of a rise in interest rates to slow down dramatically, which could produce a material extension in these securities’ duration. In other words, at these low yields, that 3.5% five-year bond could turn into a 10-year bond with a relatively small move in interest rates. As we mentioned above, the lower the coupon for a given duration, the more volatile that coupon will become and if you include negative convexity to the mix, you have the recipe to turn what is seemingly a benign situation into a very toxic environment.
As prepays slow and MBS extends so too does the need to hedge prepay risk thus turning MBS hedging Treasury buyers into sellers. The more rates rise, the more these low coupon MBS will extend and the more Treasuries they will sell. This is negative convexity in the other direction. Add to that waning Treasury demand due to a decline in foreign reserve accumulation and a market that is staring at an eventual Fed exit and you could quickly kick the bond market’s reflexive process into reverse. And that, my friends, is the anatomy of a typical boom and bust.