Cy Berlowitz asked a great question in comment form on the Blowback page, and as I finished my reply to the comment, I realized it would be good to copy it here on the main page.
Here’s Cy’s question:
Howard, I’m trying to figure out when MBS have the wind at their backs. Am I correct in thinking that the wider the spread between the 10-year bond and the 90-day T-Bill the better it is for MBS?
My response after the break…..
You are correct that the way we used to look at mortgages was that the 10-year Treasury determined the mortgage rate, and the 3-month Treasury was a good proxy for financing rates. (Pre-1993) That made the Bills/Notes spread the one to watch for the MBS carry trade.
In the 90′s things shifted. LIBOR became the determining rate for almost all debt, and the effective life of mortgages shifted lower due to introduction of ARMs, hybrid ARMs and increasing popularity of 15 year mortgages. Today we might look at the 5 year or 7 year as the proxy for mortgage rates, and 3 month LIBOR as the proxy for repo.
You could also simply look at the average conventional fixed rate, published by Freddie Mac every Thursday evening, and also look at the 1-yr ARM and 5-1 hybrid ARM rates to determine what new money invested in MBS can make on a leveraged basis.
Here’s how I would do it. First, you need to subtract out the servicing cost from the mortgage rate (50 BP’s is conservative), so a 4.75% fixed rate mortgage with 35 BP’s 3-mon L translates into 425 – 35, or 390 basis points of “raw” spread. Figure half of that disappears for robust hedging, and a third of it for being partially hedged. These days, all the amREITs hedge at least partially, so they are capturing anywhere from 200 to 250 basis points of spread after hedging.
At 6-1 leverage, that would be 12% to 15% per dollar of capital in carry profits, plus the underlying mortgage yield of 4.25%. Theoretically, then, a fully fixed rate amREIT that partially hedges could earn 19.25% before costs on each dollar of equity capital.
Figure 100 basis points of expenses for a large REIT (over $1 billion capital) and up to 200 basis points for a smaller REIT with $500 million or less in capital. That takes the ROE down to 17.25% to 18.25%. Then you need to adjust for the premium or discount to book. A stock trading at 1.1 times book would give an expected dividend yield ranging from 15.5% to 16.9%.
Running the same numbers with ARMs, fully hedged, might give you net of 200 basis points on the carry. [Note: 1-year ARMs are much cheaper to hedge, so I would use 200 BP’s net for both.] At 6-1 leverage again, that makes 1200 basis points, or 12% carry profits. Add in the 3.75% mortgage rate less 50 BP’s for servicing, and you have 15.25% raw return on capital. Subtract 100 to 200 basis points for expenses of the REIT, and you get ROE of 12.25% to 13.25%.
For ANH I would assume 125 basis points of costs, since it has about $900 million in capital. If it’s trading to a 0.9x discount to book value like ANH, divide by 0.9 to get expected investor yield of 13%/0.9, or 14.44% for new money they are putting to work today.
I just checked their website for presentations, and one dated June 2nd this year (see slide 4) is pretty much in line with my “guesstimates”.