A few months ago, ARMOUR Residential REIT (ARR) was trading around $7.00/share, but in May 2013, investors can now purchase shares at around $5.50. This underperformance is not recent. Cumulative before and after tax returns (using highest federal marginal tax rate) are 16.89% and .86% for the last two-and-a-half years. On the other hand, the S&P500 SPDR ETF (SPY) has an after tax return of 36.86%. The chart below illustrates the returns on a quarterly basis. The large difference between the before and after returns of ARR is due to the substantial monthly dividends (>10% yield) subject to ordinary income tax.
This article is very long, but will offer a comprehensive guide on how a rising interest rate environment and conflicts of interest between ARR's external manager - Armour Residential Management LLC (ARRM) and sub-manager - Staton Bell Blank Check LLC (SBBC), will reduce dividends and NAV: net asset value. I believe buying the stock with at least a 15% discount to NAV is a sufficient margin of safety. Even after investing in ARR, shareholders must continue to monitor the company and keep watch of two things: the growing preferred stock balance, which is really debt and any equity offerings below NAV.
Rising Interest Rates
Interest Rates are at historic lows because of the massive purchases of fixed income securities including residential mortgage-backed securities as part of the Federal Reserve's Quantitative Easing Program: QE. When QE inevitably comes to an end, the mortgage-backed security industry will lose a multi-billion dollar buyer. The Fed holds as of May 22, 2013 $1.178 trillion in residential mortgage-backed securities guaranteed by Fannie Mae, Freddie Mac, and Ginnie Mae (also referred to as Agency securities). This amount is a little over 10% of the entire residential (1-4 family residences plus multifamily) mortgage-backed security industry estimated at $10.789 trillion as of 12/31/2012.
As the Fed winds down purchases of mortgage-backed securities, I expect mortgage rates to climb towards, but not meet, the historical average of 8.62%. The table and chart below show historical mortgage rates. For an ARR shareholder, a slow ascent in interest rates allows the agency portfolio to reinvest prepayments at market yields to offset faster rising short term borrowing costs.
Declining Estimated Dividend Rates and Net Asset Values
The below exhibit, copied/pasted from ARR's 3/31/2013 10-Q filing, details what could happen with a 1% change in interest rates on the Net Interest Income and Portfolio Value.
I agree with the company's projections that increases in interest rates hurt shareholders both on an estimated dividend rate and net asset value basis. The rest of the article will explore the multitude of factors that drive this. For the coming future estimated dividend rate and net asset value sections, please refer to the below exhibit in which I list self-calculated and company provided statistics from the company's 10-K/Q as reference.
Estimated dividend rate will decline from increasing interest rates
Note: ARR's estimated dividend rate is a function of its net interest margin multiplied by leverage. This also can be referred to as return on equity because REITs typically distribute nearly all their earnings. There are many ways to calculate leverage, but all of them will lead to the same intuition I describe in the article, but with slightly different numbers. Please see the footnote in the Statistics and Data table above for calculation. ARR does not provide a breakdown of prepayment versus scheduled repayment rates. My use of prepayment rate will underestimate principal payback.
1. Net interest margin will decline
- Slow rising average mortgage coupons/yields: the ability to reinvest at higher coupon/yielding securities depends on principal prepayments which ranged from a 9.1% to 19.30% annual rate
- Prepayments will slow: average annualized prepayment should decrease with rising interest rates as borrowers who could refinance have already done so
- Fast rising cost of funds: Short maturity repos are repaid at a 936% to 2808% annual rate. Cost of entering into new swaps, swaptions, and futures rise as well
- Swaps, swaptions, and futures cover 40 - 60% of repos but not enough to offset rising repo costs
2. Leverage will decline
- Increased chance for margin calls: Agency securities are posted as collateral but agency security prices move lower with higher interest rates
- More collateral needed for new repo contracts: Haircut for repurchase had gone from 5.3% to 4.8% from 12/31/2010 to 3/31/2013, but it could increase again
3. Share prices will be below NAV due to the two points above, however ARR will continue to issue new equity at a brisk pace because ARRM is compensated on gross equity raised and has little stake in the company (will be explained further in the article).
In the chart below, I highlight the declining estimated dividend rate with the drop in net interest margin and leverage. From the Statistics and Data table, falling agency security yields and rising cost of funds (repos plus derivatives: swaps, swaptions, futures) drop net interest income. Since leverage is relatively stable, the primary catalyst for the drop in estimated dividend rate is falling net interest margin.
Scenario Analysis: If interest rates rise by 1%, I'm projecting the possible changes to estimated dividend rate within a month:
*I'm using 3/31/2013 data from the Statistics and Data exhibit earlier
- Agency portfolio only has enough time to reinvest its 1.31% of total portfolio (15.7% annualized prepayment) into new agency securities at an asset yield of 3.33% and the asset yield climbs from 2.33% to 2.34%
- Roughly half of the unhedged repurchase agreements (25% of total repos - weighted average maturity of 36 days) expire and are reinvested at a 1.46% higher rate. Cost of funds increase from .98% to 1.23%
- Leverage decreases from 10.50 to 10.00 due to declining asset values, margin calls, and higher haircuts (this is reasonable since 10 is low relative to the company's prior ratios see Statistics and Data exhibit)
Multiplying the new net interest margin of 1.11% (was 1.35%) by a leverage of 10.00 will equal an estimated dividend rate of 11.1%, a 21.72% decrease from 14.18% and a net interest margin decrease of 17.78%. This is lower than the 11.89% decrease in net interest income (interest income - interest expense) suggested by ARR. Although our calculation methodologies differ, we both project a large decrease in net interest whether it is related to margin or income. I do not know if ARR would factor in a decline in leverage as part of their analysis, but I hope this exercise illustrates the sensitivity of the estimated dividend rate to a small increase in interest rates.
NAV will decline from increasing interest rates
1. Agency security values will decline
2. Repo values will remain close to par: short maturity and continuous rollover at market interest rates
3. Swap, Swaption, and Futures value will increase, but not enough to cover the drop in Agency security values: Swap, Swaption, and Futures' aggregate notional value is about half of agency security's total current face value
4. Share prices will be below NAV due the three points above, however ARR will continue to issue new equity at a brisk pace because ARRM is compensated on gross equity raised and has little stake in the company (will be explained further in the article)
Scenario Analysis: If interest rates rise by 1% I'm projecting the possible changes to NAV within a month:
*I'm using 3/31/2013 data from the Statistics and Data exhibit earlier and ARR's May 9th monthly company update for duration statistics
- Agency portfolio with a duration of 3.83 will have a value decline of 3.83%
- Repurchase agreements will remain at par
- Swap, Swaptions, and Futures roughly half of agency face value with a 5.80 duration will have an increase of 2.9%
I project a .93% drop in NAV versus ARR's 2.42% drop. Some of the differences can be attributed to using the May 9th monthly company update which has different durations than what might have been used on the 03/31/2013 10-Q. The important thing is that we both agree that NAV will drop, but not as severely as the estimated dividend rate.
Conflicts of Interest: ARRM has incentives to enrich itself rather than shareholders
1. Management Fee is paid based on gross equity raised
Although the management fee as a percent of gross equity isn't high, the fee is not calculated off current equity, which provides incentives to pursue equity offerings as opposed to increasing shareholder value. This is stated in the 3/31/2013 10-Q report:
2. ARRM can raise additional equity without shareholder approval
This allows ARRM to quickly increase their management fees. The chart below shows that from the 12/31/2010 to 3/31/2013 quarter total equity increased over 2400% leading to a management fee increase of more than 2000% while the investor before tax return is only 24%.
Assuming equity is raised at or above NAV and leverage stays constant, raising common equity in a declining interest rate environment limits shareholder returns. ARR's rapid growth in equity base provides funding to purchase assets at lower yields. This limits growth in net asset value per common share and lowers net interest margin. Shareholders benefit if ARR issues equity in a rising interest environment as the fall in net asset value per common share is partially offset and net interest margin increases.
Fortunately, ARR has an incentive to time offerings close to or above NAV to maximize gross equity. I can confirm that prior equity offerings were not offered significantly below NAV. The chart below shows how Avg High/Low prices for each quarter remained relatively close to or at a noticeable premium to book value per common share.
The below chart shows the damaging effects of the equity offerings despite having them offered at or above NAV in the form of colossal prepayment figures. Agency securities invested with equity raises can be seen as a form of mortgage prepayment, as both require the capital to be reinvested at prevailing agency security yields. I can make this assumption for two reasons: leverage is relatively consistent across quarters and book value of equity and cumulative equity raised is very close. You can see how the adjusted quarterly prepayment rate is many times larger than the estimated quarter prepayment rate if ARR did not raise equity.
3. ARR board of directors are employees of ARRM or directors of a competing REIT
4 out of the 9 board of directors work for either the external manager ARRM or ARRM's sub-manager - SBBC. Unfortunately, the same 4 directors are also running JAVELIN Mortgage Investment Corp (JMI) which invests in agency and non-agency mortgage-backed securities financed by repos and hedged with derivatives. Another 4 directors are working as directors for JMI as well. I highlighted directors with significant conflicts of interest in red and directors who might have some conflicts of interest in orange in the chart below. (click to enlarge)
4. ARR executive directors and ARRM employees do not own many ARR shares
ARRM earned $6.6m in management fees in the 3/31/2013 quarter. 25% of the fee ($1.6m) is owed to SBBC as part of the sub-management agreement. As of 12/31/2013, ARRM had 14 employees, which equates to a little over $355,000/quarter or $1,400,000/year per employee. The exhibit below from the DEF 14A filed March 18, shows that Scott J. Ulm and Jeffrey J. Zimmer, with very little stake in ARR relative to the management fees, collected 130,000 - 200,000 shares times a price of $6/share is $780,000 - $1,200,000 versus the $1,400,000/year per employee calculated above. Daniel C. Staton and Marc H. Bell of SBBC have a sizable stake and have interests aligned with ARR shareholders in this respect. The 3/31/2013 10-Q mentioned an amendment to the 2009 Stock Incentive Plan increased the shares issuable from 250,000 to 2,000,000 for board members and ARRM employees. As of 3/31/2013, Unvested Awards Outstanding was 1,800,247 shares. This is a step in the right direction, but considering 2,000,000 shares at $6/share is just $12m versus $6.6m in management fees accrued each quarter.
(click to enlarge)
A downside to the REIT structure is the 5/50 rule and the 9.9% single shareholder restriction. The 5/50 rule says that five or fewer shareholders cannot own more than 50% of the company. Also any individual shareholder cannot own more than 9.9% of the company. These restrictions, despite their good intention to prevent conflicts of interest between a concentrated group of shareholders and smaller shareholders, also prevent shareholders from rallying against management conflicts of interest and abuse. I doubt shareholders would mind if certain institutional investors like BlackRock, which owns 6.16% of ARR, increases their position. BlackRock has a fiduciary duty to act on behalf of their clients and vote against shareholder destructive proposals.
5. The external management and sub-management agreement are costly to terminate
According to the amended and restated management agreement with ARRM dated June 18, 2012, the agreement cannot be cancelled without cause until June 18, 2022. Thereafter the agreement would be automatically renewed every 5 years unless there is a written termination 180 days prior. Any termination without cause after the initial 10 years would force ARR to reimburse ARRM with 3 times the management fee for the preceding 12 months. There is also an agreement with the sub-manager SBBC which gives the option to terminate the sub-management agreement on November 6, 2014, which will force ARR to pay a hefty fee of 6.16 times the annualized rate of the sub-management fee for the prior three months.
Caution: Additional borrowings disguised as Preferred Stock
One of the troubling finds on ARR's balance sheet is the use of preferred stock as debt. Under some of the company's Master Repurchase Agreements and Master Swap Agreements, there is a leverage restriction of 12 times stockholder's equity. ARR's 8.25% Series A (ARR-PA) and 7.875% Series B (ARR-PB) perpetual preferred stock pay steady dividends unless redeemed by ARR at $25/share. These preferred stocks are not convertible to common equity unless a shareholder breaks the REIT structure's 9.9% single shareholder rule. Issuing these types of preferred stock is akin to straight debt. The benefits of preferred stock is that it prevents common stock dilution and protects common share dividends, because proceeds and equity classification allow additional repurchase agreements and agency securities, which have a 14.18% estimated return of equity as of 3/31/2013. The drawbacks are the expensive dividend payments of 7.875% and 8.25% and increased leverage ratios relative to common equity. Since ARR is already highly leveraged, I'm troubled by the addition of preferred stock to the mix and hope the company thinks carefully about future offerings. The table below shows a calculation of the contractual leverage and the adjusted contractual leverage which reclassifies the preferred stock as debt.
Buy ARR only at a 15% discount to NAV
A rising interest rate environment and conflicts of interest between ARR and its external manager - ARRM and sub-manager - SBBC will decrease both estimated dividend rates and NAV, therefore, investors should purchase shares of common equity at 15% below NAV. Even after buying shares, shareholders need to monitor ARR and be on the lookout for two issues: continued growth in its preferred stock which is really masked debt and any equity issuances below NAV.