3. This is March status. I assume that the majority defaulted after reset, not before. Well, at THAT rate probably 80% will default by the end of the year

I wouldn't assume that; there's no reason to. Just take the 7% in FC/REO: those have been delinquent for at least 90 days as of March; you don't miss a payment in January and end up REO in March.

It's highly likely that a lot of those loans started missing payments almost immediately; remember all the EPDs?

I also don't think it helps to think in terms of "dropping" the payoffs. The 2/28 product was designed to mostly pay off in 2 years, and lo and behold, it did. Until the refi opportunities went away, as they will when the rate cycle goes away from you.

But you have to remember that any significantly aged vintage will end up looking like that 2006-reset pie. The issue is not that the vintages prepay; they all do. The issue is how fast (or how slow).

Lama, these charts are based on status as of the end of the first quarter of 2007.

Remember that what you are seeing is that 37% of 2/28s that were originated in very late 2004-very late 2005 have paid off by the end of the first quarter of 2007. In other words, the importance of 2007 is that this is the calendar year in which those loans have their scheduled first payment reset. So, 2/28 loans that had an actual first payment reset in Q107 had a first payment date of 1/05, 2/05, or 3/05 (and closing date of somewhere around 11/04-2/05). Those loans could have refinanced at any time from the beginning of 2005 until March of 2007. The issue, I think, is that if that 45% of "current" loans hadn't refi'd by 3/07, they're likely to have a hard time refinancing ahead of their 2007 reset, since between tightening guidelines and rising rates, they're no longer "in the money." The implication is that a big chunk of the 2005 originations "missed the boat." Since 75% of these loans hadn't reset as of the report date, you can conclude that defaults have nowhere to go but up.

Well, bear in mind that these two vintages are probably accounting for the lion's share of the modifications that have been causing so much uproar lately. A modification is never counted as a payoff, and once payments resume it is not reported as delinquent. It's quite possible that a more or less significant slice of these "current" loans in either vintage are modifications (that have not yet redefaulted).

BG, I've seen some recent roll-rate and payment velocity stuff on the 2/28s that's pretty scary. It's also 1) not on the web and 2) not free.

I'm trying to keep my eye out for anything like that to become web-available.