This was an archived piece that I had published to the wrong section that I don't think ever made it on the blog (please note that all references to MBS coupons and duration relative to Treasuries ONLY pertains to the time of the intial writing in early 2009). It may be useful as a knowledge base. Let me know if you think it needs more:
Speeds and Spreads....
Though it is very important to understand that MBS and treasuries are very different, there are important ways to analyze them together. To put it simply, Treasuries are the risk free benchmark against which numerous other fixed-income investments are compared. It's as simple as that. For the most part, treasuries are considered to be utterly risk free, whereas MBS, like other fixed-income investments have "risk." Of the greatest importance to MBS investors is NOT "credit risk," which is the risk they will not be paid back, but rather it is "call risk." In other words, there is a risk that the loans underlying MBS pools can be "called" when the homeowner sells, refinances, or undergoes another form of less savory disposition. In all cases, the investor's standpoint is the same (that's not to say that Fannie, Freddie, the servicer, and the MI company won't take a hit, but the investor is guaranteed "timely payment of principle and interest").
Because MBS have risk and treasuries don't, that which is risk free can be used as a baseline to determine just how risky anything else is. The "Risk Premium," then, is simply the difference in yield between the risk-free and the non-risk-free. This is the VERY central concept of MBS analysis known as "spread." In fact, MBS analysts are more concerned with the difference in yield between MBS and treasuries than they are with the actual yields themselves. Since there are many types of treasuries and many types of MBS, how then do we go about comparing the two in the most pertinent manner?
Say we want to start our analysis with MBS versus a 10 year treasury. In order to get a relevant basis for comparison from the MBS side, we have to find an MBS coupon that "lasts about as long" as 10 years. Remember that there is no set time limit on MBS other than the mortgage note. If people want to keep them all 30 years, they can. But of course, they don't. The question of "prepayment speed" is central to the concept of MBS analysis. This is commonly referred to as "speeds." It encompasses refinances, sales, foreclosures, etc... For mortgage brokers in particular, it's important not to confuse anything about a "prepayment penalty" with "prepayment speeds" (though penalties are gradually becoming a thing of the past it seems). So if you hear us discuss "prepays," or "speeds" or CPR (constand prepayment rate), or "weighted avergage life," or "embedded call option" (since the borrower can 'call' the note by paying it off), we are concerned with how long the average borrower in a mortgage that underlies an MBS coupon will keep paying their mortgage. One thing that might surprise some of you is that in the instance of foreclosure etc..., the secondary market investor is much less worried that the loan is defaulting because they are guaranteed "timely payment of principle and interest" and much more worried that the speeds they were anticipating are changing, so they might have cash that was locked in earning 7% all of the sudden in a market that's only paying 4.5%. Lots of money gets lost that way. Lots of banks go out of business. And we see exactly what we've seen, which is mortgage spreads blowing out far past all time wides to account for the uncertainty in prepayment speeds. To conclude, we used a baseline of 10 year treasuries agains the 4.5% coupon because the prepayment speed on a 4.5 is roughly 10 years. So to an investor choosing risk versus no risk, everything else would be apples to apples. When the yields (rates) on mortgages go down faster than treasury yields, that's what we refer to as "tightening" (because the yield curves are getting tighter to each other).