Aw, quit showboating those fancy regression analysis skills, CR: '2nd order polynomial.'

Actually, very nice work, sir!

Don't forget-- the GSEs have raised their pricing. So this is a fourth reason.

How about the possibility that the treasury rates are skewed by factors other than the economic situation? What if the mortgage rate setters are no longer are drinking the same kool-ade?

For example the inflation adjusted treasuries are predicting a long-term inflation rate of about 2-1/2% for the next couple of decades. How real is that? What kind of bet would you make on those rates?

Call the spread the "gse failure premium"--the transactions cost of turning a gse obligation into a treasury obligation when the gse defaults and has to call on the treasury to bail them out.

I wonder how closely correlated is widening of the spread with increasing oil prices, i.e., dollar devaluation.

It's all about the dollars, baby.

I'll vote for the "treasury=security blanket" hypothesis.

With inflation expectation unanchored, the fed done with cutting, this is not a good time to buy into long-term bonds.

6.32% is not a bad rate from the historical view point, no?

Yalt | 06.12.08 - 2:03 pm | #

My understanding is that the polynormal part is where the first reason comes in. When Tbill rate increases, mortgage rate should not only increase, but also increase at an accelerating rate. Thus, the spread should also increase.

Although I could be wrong on this tech explanation.

CR,
Do you really need a second order polynomial there? It looks like a pretty linear relationship. A second order polynomial will always give you a better fit (but not as good as a 50th-order polynomial!) but I bet any model selection criterion would say the linear relationship is best. That said, an analysis using a linear relationship would give you virtually the same conclusion (i.e., we're currently above the predicted rate level).

I think the anwser is that because of risk concerns, investors are paying a premium for the treasuries.

I've been holding my breath waiting for these rates to go up again for quite some time now.

All I can say is: Thank God and Keep Rising.

This housing market has a long way to go if it's going to correct back to sanity levels.

The delay tactics for the correction that we've been witnessing for the past year have done nothing but 1)suckered in a bunch more new buyers now ripe for potential defaults later on and 2) suckered in a bunch of people with real money gained from selling last year into plunking it down on yet more overpriced RE, when it could have been spent (or saved) more wisely elsewhere.

What a waste on both counts.

Politicians such as Barney Franks, Dodd, Clinton, Schumer, and now, Obama have done nothing to help the average American with their insistence on slowing down the crash. All they've done is suckered more people into the mess.

The higher these rates go to reflect the real risk that is out there, the better for all.

Watching the 10 year shoot up over the past day is awesome.

Hopefully , it'll just keep on going til there's no doubt left in anyone's mind that this RE party/puke fest is well and truly over and people can go back to doing better things with (what's left of) their money.

The answer is either simpler or harder than you people are making it.

Mortgages are priced off the swap curve. Thus, you need to use swap rates.

The next complication is that mortgages are amortizing, whereas bonds (and swaps) are bullet amortizations - i.e., principal is repaid only at maturity. This effect alone means that you cannot compare a 30-year mortgage to a 30-year swap.

Then, as CR points out, people can pre-pay their mortgages. You thus need to value a wicked path dependent option embedded in mortgages.

This then gives you the theoretical rate.

Are we done yet? No!

There is the "basis" between cash bonds and the rates implied by the hedging structure. So you thus get an "error" term (which people trade) which is the final factor determining the mortgage rate.

The biggest problems in CR's approach is that he is using Treasurys and not swaps (missing the swap spread), and the implied volatility effect (for the option). If vol goes up, the embedded option value goes up, and pushes up the mortgage rate.

But then again - it's a pretty damn good fit.

"Louis Siqueiros writes:
Instead of showing how mortgage rates compare to the the treasure 10 year note, and say they follow in in similar pattern, why not actually correctly inform us as to how mortgage rates compare to mortgage backed securities (MBS)."

That's almost straightforward. The mortgage issuer can sell "TBA" securities on a forward date (To Be Announced - the security does not exist). As mortgages are created, they are bundled and then delivered into this forward sale. (Tanta knows way more about this segment of the market than I do; mortgages are not my thing.)

Recently issued MBS with similar coupons will price similar to TBA's; but seasoned MBS may price wildly differently and so provide no guidance. This is because the underlying mortgages may have prepaid, or have different coupon rates and thus different prepayment risks.

There's also a spread between the wholesale rate and the retail rate to handle all the fees etc. embedded in the securitization. It may jump around a bit, but that volatility should be relatively small versus the raw rate risk in these products.

However, it's tough trying to get a publicly available source of MBS pricing.

risk "adverse"??? averse?