Bond Market Convexity: Objects in Mirror Are Closer Than They Appear

By Vince Foster  FEB 19, 2013 10:25 AM

It's uncertain whether we're seeing a trend change in interest rates or the continuation of the deleveraging process.


FX volatility was front and center last week as rhetoric from the G7, G20, and Bank of Japan dominated headlines and yen price action.  With US stocks grinding higher, the bond market has been bearing the brunt of the yen volatility with the US bond futures contract continuing to vibrate the 143-00 level I cited as a price objective in my year-end 2013 Bond Market Prognostication:
What Are the Levels to Watch for in 2013?
I think 143-00 is a huge number. It was a climax top in 2008. It was made support in 2011 and 2012, and it’s very close to two standard deviations in the rising channel. If violated to the downside, a bond market top could be in place, but don’t expect it to go quietly; it will likely put up a tough fight.

US Bond Futures Tick

Clearly the market has recognized the 143-00 pivot as the consolidation is in its third consecutive week and there's no doubt that a bull/bear battle is underway as participants jockey for positioning.  The key factor for me in determining who wins this battle is whether economic growth and the demand for credit is increasing, or whether recent optimism is simply a function of rising stock prices on the back of a depreciating yen.
As I wrote in Analysts Missing Key Ingrediant of Economic Growth, one of the financial metrics I follow to monitor the demand for money is the commercial banking system’s aggregate total credit assets broken down between securities and loans.  You can see on the chart below the ratio of securities held on the balance sheet to total credit assets tends to cycle with the implied carry earned by investing in mortgage-backed securities (MBS) which I use as a proxy for all depository investable assets. 
Loan-to-Deposit Vs. Securities

The loan-to-deposit ratio, which tends to run inverse to the ratio of securities, is a function of the demand for credit, and the correlation with the velocity of money (M2) demonstrates whether this credit extension (and expansion) gets turned over in the economy.  (In the aforementioned article I dig a bit deeper into this relationship citing work by Dr. Lacy Hunt and Van Hoisington of Hoisington Investment Management.)
If you think about it this makes perfect sense.  In a normal cycle when the demand for money is strong, the curve flattens from the short end in anticipation of tight monetary policy.  Banks reduce their holdings of securities due to narrowing carry rotating into loans which increases the velocity of money as credit expands in the economy.  When demand for money is soft the curve steepens in anticipation of easy monetary policy.  Banks reduce their extension of credit, rotating into securities to take advantage of wider carry, which in turn decreases the velocity of money as credit sits idle in securities.
I believe one of the primary goals of QE is to flatten the curve from the long end to reduce the carry earned on depository institutions' pool of investable assets, making them uneconomic to hold.  I wrote the following on November 19 in How QE is Impeding the Economic Recovery:
There has been much discussion on the intentions and effectiveness of QE.  It has been my contention that the main objective is not to reflate asset prices but rather to stimulate credit creation and the velocity of money.  According the Fed’s H.8 Release banks are holding over $2.6 trillion in cash that's sitting idle on their balance sheets in securities portfolios.

[Ben] Bernanke is trying to flush the banking system out of these bloated securities positions and into extending credit by lowering bond yields to levels where banks can no longer afford to hold them.  As banks replace securities with loans credit expands, the velocity of money increases which in turn will increase economic activity, employment, and corporate profitability. 
If QE has been successful in increasing the supply of credit, you would expect to see banks increasing their loan books while decreasing their securities portfolios if the demand was there.  Well, it’s not happening.  In fact, over the past year, banks have actually been increasing the ratio of securities purchased as a percent of total credit assets.
Bank Securities Vs. MBS Yield


When the Fed launched QE III in September 2012 -- which targeted virtually 100% of newly issued MBS, sending spreads and outright yields to historic lows -- you would have thought banks would have backed off purchasing securities.  However, according to the Fed’s H.8 data, between 8/29/12 and 11/28/12, total bank credit increased by $81.8 billion with securities representing a whopping $51.9 billion or 63% of the increase.  I expanded the sample size to remove any short term anomalies and found a similar trend.  In the second half of 2012, total credit assets increased by $266.2 billion with securities comprised of $132.6 billion for a still robust 50% allocation. 
I went back through 20 years of data looking at an apples-to-apples comparison for the second half of the year and there was only one six-month period that saw a greater dollar amount increase in securities purchased.  In 2002, out of the $378.7 billion increase in total credit assets, banks bought $142.6 billion in securities but representing only 38% of the increase versus a $236 billion allocation to loans.  In fact, the only years where securities comprised over 50% of total credit increase in the second half of the year were 1991, 1992, 2001, 2009, and 2010 -- basically all during or coming out of recession. To provide another comp, in the second half of 2011, when banks added a similar $266.7 billion in total credit assets, securities represented only $84 billion of the growth (32%) versus loans of $178 billion (68%).  The typical allocation breakdown during expansions in the 1990s and 2000s was closer to 80/20 loans versus securities.  Even in the second half of 2008 as the wheels were coming off, total banking credit increased by $321 billion, of which $300 billion (93%) went to loans and only $21 billion went into securities.
The point of this analysis is twofold: 
First was to demonstrate that by calculating how banks are allocating credit assets between loans and securities, it does not appear that the demand for credit is increasing, and it could actually be decelerating. 
Second is that I now want to demonstrate how these bloated securities positions on bank balance sheets represent significant systemic risk with exponential effects on bond market volatility and thus the interpretation of the market discount.
As I posited earlier this month in What’s Driving Intense Volatility in the Bond Market? one of the three possible explanations was convexity selling by MBS holders.  Banks hold a large portion of their securities assets in callable and hence negatively convex bonds.  I have been concerned about the negative convexity risk on bank balance sheets since last June when I wrote In The Parallel Universe, Credit Risk is Interest Rate Risk:
The Fed likes to extol the benefit of low interest rates on borrowers without discussing the costs to the lender/investor and it’s not just in lower yield.  The lower the coupon for a given duration, the more volatile that coupon becomes.  Currently (with virtually the entire curve sporting negative coupons) there is tremendous embedded volatility on bank balance sheets.  Add to that the fact that the banks' holdings are primarily callable and thus short convexity, and you have the recipe for a very unpleasant situation when rates start to rise. 

Convexity can be thought of as the second derivative of duration.  Bonds that are positively convex rise more in price when yields fall and fall less in price when yields rise.  Conversely, bonds that are negatively convex rise less in price when yields fall and more in price when yields rise.  This is also known as extension risk: When rates rise, what was once a short callable becomes a long bullet increasing the duration, and thus mark to market risk on a bank’s balance sheet. 

The reason objects in mirror are closer than they appear is due to convexity.  The shape of the mirror is convex and this distorts the images it reflects.  The same can be said for convexity in the bond market. which can distort the duration of your portfolio, and in a bank’s case, the balance sheet. 
If you have a negatively convex portfolio when interest rates fall, it will look to be very short duration, and when they rise it will look to be very long.  As a consequence, when interest rates fall, you may tend to buy more duration, and conversely, when rates rise, you may tend to sell more duration.  To the fixed income value investors this is completely illogical -- buying more when prices are high and selling more when prices are low.
This is the reflexive nature of the bond market in action, and if this occurs systemically at extremes, convexity hedging can magnify the impact of interest rate movement and thus the interpretation of the discount.  For example, when falling interest rates trigger more duration buying, further pushing interest rates lower, this may be incorrectly interpreted as deflationary.  When rising interest rates trigger duration selling, pushing interest rates higher, this may be incorrectly interpreted as inflationary.
As I noted, in the second half of last year banks loaded up on very low coupon negatively convex assets as interest rates fell to a record low.  Then in December, Congress failed to extend the so-called TAG FDIC insurance on unlimited amounts of non-interest bearing deposits.  The long end of the yield curve was already rising due to the weakening yen and rising stocks.  Banks had to sell securities to meet those redemptions, which is why you see a slight dip in securities holdings and even large dip in the ratio to total credit.  Because duration was already extending they likely sold more duration, which caused interest rates to rise further driving more duration extension.
US Bond Futures Weekly

This duration selling pushed the US bond futures contract right into my 143-00 objective and the market has been battling ever since.  Now the 143-00 pivot level into the rising trend line carries even more significance.  If this area fails because the demand for money is increasing then convexity selling will intensify, increasing pressure on bank balance sheets just as they will need liquidity to extend credit.  If 143-00 holds then the demand for credit isn’t increasing and we just witnessed a minor convexity blowout head-fake.  I suspect this will be reconciled in the coming weeks.
In September the Fed told the banking system they were going all in on MBS pledging to remove virtually 100% of new production from the market.  In December they added US Treasuries and said the program was open-ended until specific economic conditions are met.  Then just over a month later comments from various Fed officials suggest they are backtracking. 
As I mentioned last week in Bank of Japan Meeting: The St. Valentine's Day Massacre, noted dove Charles Evans, for whom the open-ended QE is named, sounded off on reducing the pace of asset purchases even if the unemployment rate doesn’t meet his and the FOMC’s stated objective of 6.5%.  The so-called tapering of QE as highlighted by Kevin Ferry on the Contrarian Corner and discussed by Minyanville’s Michael Sedacca was again floated last week by Jim Bullard and Sandra Pianalto.
Do these Fed officials believe that economic growth is accelerating, or are they just worried about the perception of extremely accommodative unorthodox monetary policy as stock prices blow off to new all-time highs? 
I don’t know if they are truly drinking their own Kool-Aid or whether they are simply trying to quell expectations of unlimited QE due to excessive speculation.  What I do know is that by floating the tapering idea, the Fed is in effect looking to change the rules of the game after the bets have been placed.  Due to the systemic risk embedded in the banking system’s high concentration of securities, this is a very dangerous game.  Thus far the damage has been minimal and we still don’t know if this is the beginning of a change in trend in interest rates or whether we have more wood to chop in this deleveraging process.  My hunch is it’s the latter, but a massive negative convexity blowout may be much closer than it appears.

Twitter: @exantefactor
No positions in stocks mentioned.