Posted
Based on the market's recent price action, it looks like both Wall Street and the Fed are trying to defend the market around its current levels. I'd suspect that the target level for the 10-year note is below 3.20%, consistent with the notion that 3.25% is a danger point for the market that the authorities don't want to test. The Fed has recently been in the market buying moderate amounts of 5s and 10s over the last few sessions; I'd interpret these purchases as trying to prod rates lower, as a herder prods his sheep. The Fed is capable of more brutish actions (the equivalent of a 2x4 to a donkey's head), although I suspect they'd like to limit their involvement; they'll be fighting this battle for quite a while, and likely want to husband their ammunition.
The MBA's application index has recently exhibited some quirky behavior. The index of refinancing applications declined by more than 20% last week, even as mortgage rates remained fairly stable. While the index rose slightly this week, it remains significantly lower than implied by the historical relationship between the index and rates. (At the average rate of 4.94% quoted last week by Bankrate.com, the index should have printed almost 25% higher.)
One possible factor is the integration of Countrywide's lending operations into those of Bank of America. It's possible that the reorganization (which killed the Countrywide brand) caused a change in how they report activity to the MBA, and resulted in a one-time discontinuity in the index. Therefore, I'd advise some caution in using the MBA's indices as predictors of activity levels over the next few months.
I recently wrote a column for Asset Securitization Report on financial models and risk management that can be accessed here. This is the flip side of the "greed versus stupidity" issue broached by a NY Times columnist that I focused on in an earlier column (April 7th, I believe). There has been a lot written about the problems associated with models; interested readers should probably read current books such as "The Black Swan."
The focus of my concern is not on "bad models" per se but the improper use of models. Any model or analytical system is simply a tool. While you cannot operate a securities business in the 21st century without extensive use of analytics, it's the job of managers to be sure that their firm's financial position is not stretched to the point where a major problem with a model's output would put the firm at risk. The failures over the past year can't be laid at the feet of the quants. The management of financial firms should manage and fund their positions with an eye toward how badly the firm could suffer if the models are completely wrong.
It would be a mistake for lenders and non-Wall Street participants to view this subject as being unrelated to their businesses, as it has a direct bearing on mortgage rates and lending practices. For example, the lack of confidence in credit performance models accounts in large part for why jumbo lending rates remain stubbornly high. As I've noted in previous reports, highly-rated non-agency MBS continue to trade very poorly. (AAA-rated private-label MBS backed by prime fixed-rate jumbos currently trade in the area of 25 points behind where similar-coupon Fannies are quoted; a few years ago, this bond would have traded about ¾ of a point behind Fannies.) Investors are currently pricing in the risk that credit losses on these securities will be much higher than estimated and implied by the subordination levels dictated by the ratings agencies. (This is due to the poor recent performance of the rating agencies' predictions, in the context of continued housing market weakness.)
As a result, investors are making extremely onerous assumptions (with respect to both required yields and expected prepayments and losses) for any security not carrying some form of government guarantee. The resulting lack of an economic securitization outlet requires lenders to price jumbo loans unattractively, as they have to be either held in portfolio or sold in whole loan form (a time-consuming and difficult process).
A related discussion (and personal pet peeve) is the whole issue of "stupidity," in part because I'm not in the habit of referring to people with math and physics PhDs as "stupid." However, just because someone is a math genius doesn't mean that he understands economics and human behavior. My favorite story along those lines was a conversation that supposedly took place in 1994, when rates were rising sharply and a number of hedge funds failed. The story was that the manager of a fund was lamenting its failure to a friend over drinks. The conversation went something like this:
"I can't believe our hedges didn't work. We have the smartest guys working for us. Our head of analytics used to be a scientist in the Soviet space program!"
"Three years ago, that guy was a Communist."
While probably apocryphal, that conversation is instructive. In addressing the factors that caused the financial crisis, we should remember that modelers are not as good at projecting the collective behavior of large numbers of people over time as they are in calculating the movement of particles or the accuracy of ballistic missiles. In many cases, the problems were not caused by a lack of intelligence as much as arrogance and the inability to understand the limitations of quantitative analysis.
Secondary Marketing Managers and Capital Markets Desks, if you are interested in subscribing to the same fixed income and mortgage market data we use:
.