American Capital Agency Corp. (AGNC) is a mortgage REIT that invests primarily in Agency RMBS. Its goal is to preserve net asset value, while generating attractive risk-adjusted returns for distribution to its stockholders via regular quarterly dividends. It does this through a combination of net interest income and net realized gains and losses on its investments and hedging activities. It funds itself primarily through repurchase agreements (with a base of capital from shareholders) and lately partially through the dollar roll market. AGNC is externally managed and advised by American Capital AGNC Management LLC, which is a wholly owned subsidiary of American Capital (ACAS) -- a publicly traded private equity firm and global asset manager.
AGNC took a huge comprehensive income loss of $1.57 per common share in Q1 2013 as mortgage rates rose slightly. The book value fell from $31.64 per common share on Dec. 31, 2012, to $28.93 on March 31, 2013 -- a loss of -$2.71 per share. This happened while the underlying 30-year U.S. Treasury Bond yield and the national average 30-year fixed-rate mortgage rates were only slightly changed during the quarter.
In AGNC's case, it buys "special" low CPR (constant prepayment rate) agency RMBS. In this way, it hopes to get better returns by having low CPR fluctuations. However, the opposite happened in Q1 2013. AGNC's agency RMBS lost some of then extra premium for their specialness in Q1 2013 due to mortgage rates going up. The following table shows the before and after values for AGNC's agency RMBS.
Click to enlarge images.
As one can see from the table, the 30-year fixed-rate 3.0% agency RMBS fell from $104.84 to $103.11 from Dec. 31, 2012, to March 31, 2013. The 30-year fixed-rate 3.5% agency RMBS fell from $106.66 to $105.58. The 30-year fixed-rate 4.0% agency RMBS fell from even less from $107.22 to $106.61. These don't seem that bad until you consider the leverage involved. This was 8.2x on average during Q1 2013, including TBA positions; leverage was 8.1x as of March 31, 2013. This can multiply that 1%+ figure to 8%+ for book value losses. To get more exact figures, you have to use the actual numbers of each type of RMBS to do calculations. However, this is a good ballpark example.
AGNC hedges rate increases primarily with interest rate swaps and "swaptions" (options to enter into interest rate swaps). The problem in Q1 2013 was that the interest rates did not change that much. A lot of the premium on the agency RMBS that was lost was due to the thought that the Fed might begin tapering QE soon. Apparently the expectation of mortgage rate hikes was enough to decrease the premium on the agency RMBS, especially the "special" agency RMBS that AGNC buys. If you look at the table of the swaps values during the period (see the top of the table above), you can see that the changes in the value of the swaps don't begin to cover the losses in the value of the agency RMBS.
Part of the reason for this is that AGNC does not try to completely cover the possible interest rate related losses with swaps and swaptions. Part of the reason is that there were actually only little changes to interest rates in Q1 2013. Therefore, the interest rate swaps could not be effective in defraying the losses. Finally, AGNC's TBA operations (dollar roll operations) were largely uncovered by swaps and swaptions during Q1 2013, although AGNC changed this for Q2 2013.
How is Q2 going to compare? The following recent MBS pricing table gives a relative idea:
Admittedly, the above MBS values are not those of AGNC. However, they give an approximate picture of what has happened so far in Q2 2013. Also if you look at the values in the previous table above from AGNC's Q1 2012, you can see that these values are grossly different from those. One might therefore conclude that the losses from the above changes (without considering swaps and swaptions) will represent a huge loss in book value to AGNC in Q2 2013. Of course, this presumes that Q2 will end in approximately the current values.
According to the chart, the 30-year fixed-rate FNMA 3.0% RMBS is now about 3% below its value at the beginning of Q2 2013. Other losses were substantially less. Still one might ballpark that the losses might be closer to 2% in Q2 vs. 1% in Q1 2013. When you figure in leverage, that would bring the book value loss closer to 16%. Also keep in mind that June 7, 2013, was a very bad day for the 30-year U.S. Treasury bond. The yield closed at 3.33% from 3.21% early in the day.
Some of the above losses would be expected to be defrayed by interest rates. However, the interest rates have not moved in sync with the agency fixed-rate mortgage rates in Q2 2013. In fact the mortgage rates, which had been pricing at an approximate 50 bps premium to the 30-year U.S. Treasury bond yield, now seem to be pricing at approximately a 75+ bps premium to that yield. This means that approximately 25+ bps of the mortgage rate change will not be covered by swaps or swaptions at all because there was no simultaneous interest rate change.
The Freddie Mac table below of the 2013 monthly mortgage rates shows just how relatively big the mortgage rate change has been.
The changes in Q2 2013 are clearly much bigger than the changes in Q1 2013. Plus, the data does not really yet reflect the large rise in June 2013 yet. For June 3-7, 2013, the FRM 30-year fixed-mortgage rate averaged approximately 4.15%. This is another leap up from former values. It is far up from the 3.55% averaged in March 2013 and the 3.54% averaged in May 2013. Rates started this almost meteoric rise in mid-May 2013, and they do not seem to have topped out yet, although this week may see a topping process.
According to the June 5, 2013, Mortgage Rate Trend Survey, 36% of those surveyed think mortgage rates will remain unchanged over the next 30 days, 32% think they will move up slightly, 28% think they will move down slightly, and 4% think they will rise significantly. In other words, many think we will end Q2 with mortgage rates approximately where they ended Wednesday, June 5, 2013, which was at 4.17%. If they did so, this would be an approximate 0.60% rise from the average value in March 2013 of 3.55%. I can't easily get an accurate calculation without much more detailed information (and it might change drastically even if I did), but it seems likely that the book value loss will be in excess of 15%. Remember the approximately 25 bps of mortgage rate change that have not been matched by long- or short-term bond yield changes. Actually, for this calculation the shorter-term bond yields may be more important. The swaps and swaptions are generally based on them. For readers' own use, the 10-year yield today, June 10, 2013, is 2.20%. The five-year yield is 1.12%. The two-year yield is 0.31%. The 30-year yield is 3.34%. Each yield is up today. Readers can compare these to the values in the table above for AGNC's Q1 2013.
In sum, the picture for AGNC book value losses for Q2 2013 is ugly. It may improve by the end of the quarter, but the recent survey data indicates flat performance instead. It seems hard to believe, but AGNC may lose roughly $5 per share in book value in Q2 2013. This would bring the book value down from $28.93 at the end of Q1 2013 to roughly $23.93 at the end of Q2 2013.
I would note that this is a very rough ballpark estimate. I arrived at this estimate without much of the exact data that the company has available to it. It is not meant to be precise. It is only meant to provide a ballpark warning for investors. Q2 2013 book value losses at this data point stand to be very bad indeed. This does not mean that AGNC is a bad stock. I generally like AGNC and its management team. However, it may be wise to wait until after the Q2 report before buying AGNC. Other primarily agency RMBS investing mortgage REITs are likely in roughly the same boat. A couple of these are Annaly Capital Management (NLY) and Western Asset Mortgage Capital Corp. (WMC). They should both suffer some of the same bruises.
All of these beaten-down companies have great dividends: AGNC at 19.6%, WMC at 20.6%, and NLY at 13.3%. Many income investors will want to own these stocks for their dividends. Plus, one should keep in mind that those dividends are likely to go up in Q3 and beyond. The increased mortgage rates will likely lead to increased interest rate spreads for these companies. This will mean more net interest income. This is the main basis for the dividends. Thus, the situation above is very much a double-edged sword. You probably want to be on the sidelines for these until the picture becomes a bit clearer. Active traders may want to try to catch the bottom bounce, but it is hard to say exactly when this will happen.
The two-year chart of AGNC provides some technical direction for this trade:
The slow stochastic sub chart shows that AGNC is oversold. The main chart shows that AGNC is in a long-term downtrend. The price line is far below both the 50-day SMA and the 200-day SMA. The 50-day SMA is below the 200-day SMA. AGNC could be ready for a bounce, but I am not sure I would have enough guts to play it.
Other more diversified mortgage REITs with non-agency RMBS and CMBS seem like a much better bets than primarily agency RMBS investors at this time. One of those is INVESCO Mortgage Capital, Inc. (IVR). Its chart looks much better, and there is good reason for that. If you are going to try to catch a bounce, it might be appropriate to try to do it in a more diversified mortgage REIT such as IVR. This kind of company buys non-agency RMBS and CMBS (in addition to agency RMBS) at huge discounts to face value. These tend to go up in value as the real estate market's liquidity increases, the prices increase, and the mortgage rates increase.
Again, this article is only meant to ballpark the problem. None of the calculations should be taken as gospel. However, I hope they are helpful.
Note: Some of the above fundamental financial data is from Yahoo Finance.