It is time for a reality check on my ultimate subprime loss projections of approximately $300 billion (assumptions: $1.5 trillion subprime market; 30% ultimate default rate on recent multi-year originations ending in early 2007; 30% average recovery rate), particularly since my loss projections have been higher than anything coming out of the U.S. government or Wall Street. My ultimate default rate looks highly probable, and my average recovery rate may be too high.

[Note: I am not including another $1 billion of loans at risk including Alt-A ARMs and IOs, option ARMs to all types of borrowers, and “piggy-back” loans – capped-out second liens.]

Last week I met with a major mortgage servicer of geographically diverse U.S. subprime loans. They work 13 hour days trying to salvage what they can doing anything to avoid reporting a delinquency or foreclosure. They disclosed disturbing information unavailable even on trustee reports. The servicer asserted the rating agencies are incorrect in their optimism; recovery rates of 60% are unattainable. My average recovery rate assumption of 30% is also currently unattainable.

At times, the servicer has been selling loans for pennies (3 cents to 6 cents) on the dollar.

The servicer aggressively “reages” mortgages restructuring loans doomed to fail in the future. More benignly, they allow skipped payments failing to report them as delinquencies. They are doing everything they can to avoid foreclosure, because the recovery rates are dismal.

Contributing factors to the dismal recovery rates are the fact that legal costs relative to the relatively low loan balances are huge, and delays are long. Some of the assets include trailers, mobile homes and homes in areas where prices have plummeted. Some of the loans have negative equity at the time of delinquency.

The day a homeowner misses a payment, the servicer gets on the phone trying to work out a new deal. The servicer has found that if homeowners miss two payments, the loan is virtually doomed to default (in their experience), because most homeowners give up. The servicer avoids reporting delinquencies, which are usually reported one month behind prime mortgages already. If this practice is typical, the current scope of the subprime problem is underreported.

Most of the “early post-signing” defaults in loans originated through the early months of 2007, have been on stated income loans especially with loan to value ratios approaching 100%, whether they were “subprime” or not. This suggests “stated” income was overstated. Future defaults will kick in as resets on coupons occur in a soft housing market.

In the early years of a typical long-dated mortgage loan, little equity build-up occurs, and some of the subprime mortgage products have even less equity accumulation, or worse, negative equity. Some ARM products are 40-year loans with very little equity build-up in the early years. Some of the option ARMs have increasing principal balances. With housing prices stagnating and falling, the homeowner often has negative equity.

As ARM coupons reset to much higher rates, we can expect a wave of defaults combined with dismal recovery rates.

Most current models rely on purchased prepayment data and trustees’ reports. They have limited utility for marking portfolios to market or projecting ultimate losses. Using them will grossly underestimate potential losses. As market participants catch on, inventory will stagnate, and wasting assets will cause further write-downs at financial institutions with remaining subprime inventories.