Armour Residential REIT (ARR) and Annaly Capital Management (NLY) invest primarily in agency securities of which the principal and interest are backed by the US government (e.g., Freddie Mac (FMCC.OB), Fannie Mae (FNMA.OB), etc.). The profitability of both ARR and NLY are in large part tied to the shape of the yield curve. These companies borrow short term and lend long-term. As the difference between the long-term and short-term US treasury interest rate increases, the net interest margin for the company increases, which in turn leads to larger profits. Since both these companies are of REIT status, they are required to pay out 90% of their income as dividends. Consequently, the dividend payout is directly tied to the shape of the US treasury yield curve.
Using this piece of information, one can derive a simple model using the current quarter's average US treasury yield curve shape to estimate the quarter's dividend payout. The models for both NLY and ARR are shown to provide skillful estimates of the subsequent quarter's dividend. However, the model does perform better for NLY when compared with ARR. It is hypothesized that this result is due to the higher susceptibility of ARR to pre-payment risk.
In a previous article, the relationship between the US treasury yield curve, NLY's net interest spread, and NLY's quarterly dividend were investigated. Strong correlations were observed across all three variables. Further, it was found that the difference between the 10-year and 1-year US treasury yield curve was the best predictor of NLY's quarterly dividend.
A simple linear regression model is used to estimate NLY's quarterly dividend. Using daily US treasury yield data from 2006 to Q1 2013 along with NLY's corresponding quarterly dividend history over the same period (from Yahoo finance), a line of best fit was computed as seen in the Figure below. Each data point represents a single quarter's yield curve difference-dividend payout pair. The correlation between the 2 variables is extremely strong with a value of 0.95 (1.0 indicates perfect correlation). The relationship between the quarterly dividend (dependent variable, y) and yield curve difference (independent variable, x) is: y=0.22+0.15*x. A few comments on this:
- back in 2006 the yield curve difference was close 0; as such, there are a number of data points to support the y-intercept value of 0.22
- every one-point gain in interest rate difference increases the quarterly dividend by 15 cents
- Using the Q1 2013 average yield difference of 1.8, the model predicts a dividend of $0.49 per share. This is very close to recently announced dividend of $0.45.
The identical procedure was carried out using dividend data from ARR. Since ARR is a newer company, dividend data only goes back to 2009. The fitted line is shown in the Figure below. The correlation is high, with a value of 0.85, albeit not as good as the correlation observed with NLY. The formula for the fit is: y=0.2+0.06*x. A few comments on this relationship:
- the relationship between treasury yield and quarterly dividend is not as strong as with NLY (slope of 0.06 vs. slope of 0.16)
- their is no historical data for how ARR performs in a flatter yield curve environment
- the Q1 2013 data point concerns me - the treasury yield difference has actually increased some relative to Q3 2012 and Q 4 2012; however, the dividend continues to be reduced
Buffett said "never invest in a business you don't understand". At the end of the day, most agency mREITs' profitability are tied to the yield curve shape. I feel comfortable in owning mREITs because the yield curve shape can be tracked on a daily/quarterly basis, and ultimately be tied backed to the company's quarterly dividend. My mREIT holdings are overweight in NLY relative to ARR given their long history of profitability in various yield curve environments, and the excellent correlation between treasury yield and the quarterly dividend. ARR does worry me some with its recent dividend decreases, especially Q1/Q2 2013. Some of this is likely driven by pre-payment risk, which in my opinion, is tougher to relate back to the dividend. Regardless, I can sleep well at night with my mREIT holdings given the fundamental understanding of the profitability driver behind these holdings.