The author introduces an approach to valuing and analyzing agency mortgage real estate investment trusts (REITs) using various fixed-income modeling techniques, and he shows how certain methods specific to bond and derivatives markets could be applied to mortgage REITs, particularly for risk analysis. The author’s findings allow for a greater understanding of the key factors affecting agency mortgage REIT performance.

The research is useful because it provides REIT market participants with an explanation of the various sources of REIT—particularly agency mortgage REIT—risk and return. Agency mortgage REITs are an interesting asset class because their assets and liabilities are very liquid and can be accurately and fairly valued. Agency mortgage REIT stocks generally offer dividend yields relative to book value in the high single digits to low teens. With LIBOR essentially at zero and the 10-year Treasury under 3%, such high yields are rare in today’s environment.

The author attempts to discover the driving factors of REIT performance attribution and finds that agency mortgage REITs are not a play on the shape of the yield curve, even though other research has suggested otherwise. Moreover, the residual duration taken on by REITs brings a sizable share of overall returns. Although negative convexity risk is not a large driver of performance, the volatility risk premium does make a major contribution to mortgage REIT returns. Moreover, an agency mortgage REIT’s stock price is usually not close to the REIT’s book value per share. The discount or premium embedded in stock prices reflects a particular trust or distrust in management, as well as some anticipation of changes in market conditions.

The author builds a portfolio representative of a REIT’s assets by identifying a distribution of weights among a set of representative positions so that the aggregate characteristics of the representative portfolio are as similar as possible to the reported aggregate characteristics. He then defines a value function (i.e., the trade-off between being a little closer for one aggregate versus being a little closer for another aggregate). Finding the optimal representative portfolio requires minimizing this value function. The author then uses a Heath–Jarrow–Morton two-factor model as the core interest rate dynamics for derivatives and MBS valuation. The dynamics are calibrated to LIBOR and swap rates, as well as at-the-money swaption prices.

To properly consider cash flow projections, the author uses the representative item’s cohort-level speed to derive the cash flows. He also projects book value per share because these numbers are examined closely in the equity market so that a market premium or discount over book value can be derived. The author assumes that REITs do not actively trade their positions, do not rehedge, and do reinvest all asset cash flows into the same type of product. He further assumes that management and operations expenses are 1.5% of equity on an annualized basis.

The author provides unique analysis surrounding the performance attribution of an agency mortgage REIT. Further research could address a more precise modeling of mortgage REITs’ equity issuance and buyback strategy and the effect on performance. I also hope to see further research that discovers and analyzes additional reasons—apart from management perception and investor expectations—why some REITs trade at different premiums than others.