Did the Agencies Whack the Bond Market?

The price action in the bond market last week is worth re-looking at. We had a sharp run up in yields through Wednesday. That was followed by a pretty hefty correction through Friday. The obvious conclusion was that supply was the problem. “Too much Treasury paper”, said all of the media. No doubt but that supply was a factor. It is rarely as simple as it appears.

The bond dealers have a very good handle on what is coming from Treasury. They also are very aware of the demand side for these big auctions. I thought that the 2’s, 5’s and 7’s that were sold last week were well received given the size. If there is a supply ‘problem’, it was not evident in last week’s auctions. It is possible that the hiccup in the market last week was caused by something other than Treasury issuance. I think it was one of the Agencies. I believe that either FNM or FRE bought derivative contracts that created convexity on Monday, Tuesday and Wednesday. As a consequence the Street had to push more duration risk into the market. That was taking place while the Treasury was trying place $100 billion of coupons. It created the ‘fast’ market we saw last week.

The foregoing is speculation on my part. I have been observing the Agencies for many years. So call it an informed guess. I will say that the Agencies have done this type of market activity dozens of times in the past ten years. The market and business conditions that have forced them into the derivatives market in the past are aligned today. If the world were a different place and Fannie and Freddie were still private sector companies I would have bet a pretty penny that it was them who mucked up the bond market last week. The fact that they are today wards of the State, and they are still mucking things up comes as a surprise.

It is my contention that the Agencies have been mis-pricing mortgages every day, week, month and year for a very long time. I believe that they are well aware of that fact. The problems facing the Agencies today are credit related. They made loans that soured. Another reason for the failure of Fannie and Freddie is that they are derivative time bombs.

Take a plain vanilla mortgage. A 30-year with a fixed rate of 5%. Imbedded in this mortgage is an option to prepay in full at any time without a penalty. That option is potentially worth a great deal to the borrower. If interest rates rise to 6% that borrower is less likely to prepay that mortgage. However if interest rates fall to 4% that borrower is very likely to prepay because they can REFI and achieve significant benefits. The borrower can ‘win’ but they can’t ‘lose’. That is the definition of a long option position. The Agencies take all of the risk of that option but earn no premium income for it. They are writing free puts on the bond market as a result. The funded book of the Agencies is in excess of $1 trillion. These big numbers create very significant derivate risk. It is a very difficult and expensive risk to manage.

The average life of a mortgage pool increases when interest rates rise. Conversely, the average life falls when interest rates decline. The Agencies report and manage this risk. FNM defines this as their Duration Gap. Their stated objective is to have the estimated average life of their mortgage portfolio to be equal to the average life of their liabilities. They do a very good job at ‘managing’ this number. The reported results are generally in the 0-2 month level. From time to time, depending on market conditions, it will widen to 3 months. It is not often that the GAP goes to 4 months. In the past they managed the risk using derivative contracts when market conditions caused the Duration Gap to go beyond 3 months. They almost always do it before month end because they love to report numbers that look close 0. There is a degree of logic to this hedging strategy if you are a public company. It makes no sense if you are in receivership.

The following is a graph of ten-year interest rates over the past 9 months. The numbers I inserted over the vertical lines are the reported Duration Gap for Fannie Mae. Their most recent report provides March data. Note the sharp drop in yields from October through November. In that time period the FNM funding gap went from +2 to 0. The fall in rates shortened the average life by a few months. Intuitively one would think that the change in duration would be much larger given the very sharp drop in rates. However, the smart folks at Fannie know how to extend their average life. They write more 30-year mortgages. That automatically extends average life regardless of how much interest rates fall. FNM increased their Gross Mortgage Portfolio 28% and 9% on an annualized basis in the October/November 2008 period.


It is a horse of a different color when interest rates rise. Look at the relative change in numbers from February to March. The duration gap changed by five months. A large relative change. Writing new mortgages only compounds the problem. Sure enough, Fannie reduced its book of mortgages in March by 1.3% on annualized basis. We have not seen reductions in the Gross Book in some time. They are clearly continuing to manage the Duration Gap. In March they were being stingy about mortgage creation to minimize the risk of rising rates.

From March onward it is guess work. There are no FNM numbers to work with. The line for the Treasury yield has gone significantly higher since March. The impact to Fannie would be that their average life was being extended quickly. The duration gap probably widened to a level that was larger than Fannie wanted to report. Their response was to buy puts on the long bond (or a like derivative instrument). This creates the desired convexity. It also cost the taxpayers a fair bit more to sell those bonds last week. On the flip side, the dealer community must have made a bundle.

If the April report shows a gap of –1, we will know that it was FNM that sought rate protection early last week. In the scheme of things it is just noise. Last weeks bond market action is just old news.

A broader issue that should be considered is the interest rate risk practices of the Agencies. They have a $1 trillion book that they are earning less than one percent on. That is a very poor result when the yield curve is so steep. The US Treasury has 11 trillion of debt outstanding. The average life is currently 48 months and 1/3 is funded short term. If the Agency funding mimicked Treasuries it would result in incremental revenue from the mortgage portfolio of $100 – 150 billion per year. That money could come in handy given the credit losses that they face. I can hear Mr. Lockhart of FHFA saying, “Oh no, changing our duration limits would expose us to risk!” My response to that is, “Mr. Lockhart, if the ten year bond goes beyond 5% the Agencies are dead anyway so you might as well go for the ride.”

Agency mortgages should have a prepayment penalty attached to them. This would slow the churn, reduce the convexity problem and offset some of the expense of managing the prepayment risk. That will never happen unless someone new takes over at the top of this mess. Prepayment penalties would make Agency mortgages ‘less attractive’. The current leadership of the FHFA wants the government to be 80-90% of the mortgage market forever. That is a terrible long-term plan.