A Closer Look at American Capital Agency’s Cash Flows as of 2Q12

This article was submitted by Ron Hiram of Wise Analysis using our Trefis Contributors tool.

My prior articles focused on master limited partnerships (“MLPs”), an area I have long followed and invested in. My concern with overly concentrating my portfolio in MLPs has led me to examine mortgage Real Estate Investment Trusts (“mREITs”) as an alternative yield producing vehicle. Indeed, the current dividend yields on some mREITs exceed the distribution yields on many MLPs including El Paso Pipeline Partners (EPB), Enterprise Products Partners (EPD), Energy Transfer Partners (ETP), Kinder Morgan Energy Partners (KMP), Plains All American Pipeline (PAA), and Williams Partners (WPZ).

I have been evaluating Annaly Capital Management, Inc. (NLY) and American Capital Agency Corp. (AGNC). This report focuses on AGNC, a Nasdaq-listed mortgage real estate investment trust (“mREIT”) with a market capitalization of ~$12 billion and assets on the balance sheet as of 6/30/12 totaling ~$85 billion. AGNC owns, manages, and finances a portfolio of real estate related investments, including mortgage pass-through certificates, collateralized mortgage obligations, callable debentures and other securities backed by pools of mortgage loans.

Total returns generated from the date indicated through 9/9/12 (based on the $35.02 closing price) are summarized in the table below (note that the percent total return is the return for the entire period, not per annum):

Table 1

Past returns appear to be attractive, even more so than those generated by Annaly Capital Management, Inc. (NLY), another large mREIT whose 2Q12 results I reviewed in an article dated August 27, 2012. The current yield is also enticing at 14.3%. However, investors familiar with my approach know the first question I ask is what portion, if any, of the dividends I am receiving are really “earned”. I am leery of investing in entities (publicly traded partnerships or companies) that pay dividends, or fund distributions, by issuing debt or additional equity. In taking a closer look at AGNC I encountered difficulties similar to those I faced when reviewing the performance of master limited partnerships (“MLPs”). Since money is fungible and the AGNC annual report runs over 100 pages that are frequently hard to understand, ascertaining whether you are genuinely receiving a yield on your money (rather than a return of your money) can be a complicated endeavor.

Several examples can illustrate the complexities. The bulk of AGNC’s assets consist of mortgage-backed securities and debentures issued by Fannie Mae, Freddie Mac or Ginnie Mae, and of corporate debt (together, “Agency Securities”). These are classified for accounting purposes as available-for-sale and are reported at fair value with unrealized gains and losses excluded from earnings and reported as a separate component of stockholders’ equity. Another example is discontinuing hedge accounting for its interest rate swaps beginning in 4Q 2012. This will reduce the balance of net losses accumulated on the balance sheet (under “Accumulated Other Comprehensive Income”) with respect to such interest rate swaps and increase the interest expenses over the remaining contractual terms of these swaps.

Using the same definition of distributable cash flow (“DCF”) I applied to the analysis of NLY, I create a quantitative standard that I view as an indicator of AGNC’s ability to generate cash flow at a level that can sustain or support an increase in quarterly distribution rates. The definition is relatively simple: net income + amortization (a non-cash item), + losses (or minus gains) on assets & liabilities (also non-cash items), less cash used for working capital.

The results for the past 3 years are outlined in Table 2 below:

Table 2: Figures in $ Millions except coverage ratios

Results for 2Q 2012, 2Q 2011, the first half of 2012 and of 2011 (1H12 and 1H11) are outlined in Table 3 below:

Table 3: Figures in $ Millions except coverage ratios

Amortization charges reflect the purchase of Agency Securities at a premium (so valued because their stated coupon exceeds market rates). The premium is paid with the expectation that the higher than market coupon will be paid over the expected life of the security. If the expectation turns out to be wrong and the actual life is shortened due to more rapid than anticipated repayment of principal, DCF will suffer. Amortization for an mREIT is therefore a far less predictable and stable component of DCF than an MLP’s depreciation charge.

Losses (gains) on assets and liabilities in Tables 2 and 3 are comprised of realized and unrealized gains and losses on mortgage backed securities, debentures, equity securities, interest rate swaps. The losses and gains are added and subtracted, respectively, because they are non-cash items.

As part of my analysis, I also created a simplified cash flow statement designed to shed light on the sustainability of the dividends by, for example, grouping together and netting out numerous line items that deal with gains and losses that are reported in the income statement but are non-cash items (and therefore reversed out in the cash flow statement). I also separate cash generation from cash consumption and group together and net out numerous line items that deal with cash outflows for assets (e.g., acquiring assets outright or receiving assets as collateral and lending against them) and cash generated by assets (e.g., selling assets outright or giving assets as collateral and borrowing against them). This reduces AGNC’s cash flow statement to just a few lines. Results for the past 4 years are outlined in Table 4 below:

Simplified Cash Flow Statement:

Table 4: Figures in $ Millions

In these simplified cash flow statements proceeds from, and payments for, assets contain numerous types of netted items, including: a) repos and reverse repos; b) securities borrowed and loaned; c) securities purchased and sold; d) principal payments on, or maturities of, securities owned; e) equity investments (including investments in affiliates). Of course, the net increase (decrease) in cash and cash equivalents ties to the company’s financial statements.

Results for 2Q 2012, 2Q 2011, 1H12 and 1H11 are outlined in Table 5:

Table 5: Figures in $ Millions

Roughly speaking, on a net basis over the 3-year period of 2009-2011, AGNC generated ~$1.3 billion, raised a further $5.7 billion by issuing equity and non-repo debt (of which only $~54 million was from such debt) and used the total of ~$7.0 billion to increase its portfolio ($4.5 billion), to pay dividends ($0.9 billion) and to increase its cash and restricted cash balances ($1.6 billion). Over this period, AGNC has demonstrated an ability to generate cash sufficient to both fund dividends and supplement funds raised via issuance of equity and debt in order to increase the size of the investment portfolio.

For the 6-month period ending 6/30/12, AGNC generated ~$1.1 billion and raised ~ $3.4 billion via issuance of common shares, preferred shares, and debt. It used the total of $4.5 billion to increase its portfolio (~$3.2 billion), to pay dividends (~$0.6 billion), and to increase its cash balance ($0.7 billion).

Clearly a 14.3% yield does not come without risks and past performance may not be a good indicator of future performance. All mREITs, including AGNC, have benefited from the accommodative stance of the Federal Reserve Bank which, for the past several years, has resulted in a relatively steep yield curve, albeit at low absolute rates. The shape of yield curve and amount of leverage (the bulk of which is generated via the repurchase markets) are the key drivers of return for AGNC which relies primarily on short-term borrowings to acquire Agency Securities with long-term maturities. Accordingly, profitability may be adversely affected if short-term interest rates increase. They could, as noted, also be adversely affected if mortgage rates decrease since this is likely to result in more rapid prepayments. In fact, in its 2011 annual report AGNC estimated that a 1% increase in rates is likely to cause far less damage (3.1% drop in projected net interest income) than a 1% decrease in interest rates (13.2% drop in projected net interest income). AGNC’s Form 10-K lists numerous other risks.

I look at several risk and performance parameters, including:

Leverage: In reviewing the table below, note that it is based on my more simplistic calculations of leverage (management calculates leverage at period end by dividing the sum of the amount outstanding under repurchase agreements, net receivable / payable for unsettled agency securities and other debt by total stockholders’ equity at period end). In 2Q12 management decreased leverage, in part through asset sales, and also repositioned the portfolio into lower coupon agency Mortgage Backed Securities (“MBS”) and lower loan balance and Home Affordable Refinance Program (“HARP”) securities, which are less susceptible to prepayment risk, reducing the impact of the decline in long-term interest rates on the portfolio. Reduced leverage reduces both interest rate risk and systemic risk (e.g., crisis in Europe, regulatory pressures for mortgage finance reform, future of Freddie Mac & Fannie Mae, SEC review of the exemption granted to mortgage REITs from the 1940 Act which would cause them to be considered as mutual funds). However, in 2012 leverage has increased:

Sensitivity to changes in interest rates: data provided by management is outlined in Table 6 below. It indicates a limited exposure to changes in interest rates as of 6/30/12:

1. Assuming movement is in the entire yield curve
2. Including interest expense on interest rate swaps, but excluding costs of supplemental hedges

Table 6

However, this may not be a reliable indication of exposure to changes in interest rates for two major reasons: first, the assumption that the entire yield curve moves in tandem is not as bad a scenario for AGNC as a further flattening (to say nothing of inversion) of the curve. Second, management’s estimates speed of prepayment vary at each interest rate level and we have no way of knowing how conservative or aggressive these assumptions are.

Prepayment speeds as reflected by the Constant Prepayment Rate (“CPR”): In the aggregate, mortgage backed securities purchased by AGNC at a premium exceed those it purchased at a discount. Therefore, the faster the prepayment rate the greater the loss. AGNC’s 2011 Annual Report presents the following illustration of how significant is the impact of CPR rates on Return on Equity (ROE):

Table 7

In addition to the adverse effect on ROE, the portfolio itself could sustain losses as a result of faster prepayments. The 2Q12 report indicates AGNC holds its portfolio of Agency MBS at a ~4.7% premium to the face value of the mortgages and that there has been a further mark-to-market (“MTM”) appreciation of ~2.2% for a total of ~7% excess over the adjusted (i.e., unamortized) purchase price. As I see it, if prepayments accelerate, the 4.7% premium will contract and the MTM gains will also decline. Hedges, depending on how effective and extensive they are, can help offset the potential loss which, on a $76 billion portfolio, could be very significant. A rear-view mirror of CPR rates indicates they have been holding steady for the past year:

Duration: Duration is the length of time required to recoup losses caused by a percent increase in short and long-term interest rates (losses are recouped by reinvesting at higher interest rates). Unfortunately, AGNC does not provide duration data.

Net interest spread: data provided by management indicates net interest spread has been declining since 6/30/11:

Management noted in its 2Q12 report that it had repositioned the portfolio into lower coupon agency MBS to further protect our portfolio against prepayment risk. The weighted average agency MBS coupon was 3.86% as of June 30, 2012 compared to 4.23% as of December 31, 2011.

Book value per share:

In summary, the bulk of investor returns have come from share price appreciation rather than dividends. Looking ahead, I don’t expect this to continue and believe investors should not factor in future share price appreciation. change. If management’s measurement of net interest margin is correct, the current yield on average interest earning assets (2.73%), net interest margin (1.65%) and leverage (~8) indicate a return on equity (and hence sustainable distributions) of about 14.3%, which equals the current yield. This very rough, back-of-the-envelope, calculation indicates that distributions are sustainable under current conditions. My assessment is that there is less risk in NLY than in AGNC because its leverage is lower, it appears to be less adversely affected by changes in interest rates, and it trades closer to book value (possibly offering more protection against share price declines). On the other hand, AGNC has demonstrated, and appears to continue to offer, better returns. I cannot make a precise risk-reward calculation and would be comfortable investing in either (bearing in mind their returns will be highly correlated). As always, there is a risk that current conditions will change and investors should perform their own due diligence and assess their individual tolerance for risk before buying or selling the shares.