Last week Countrywide Credit Industries announced that it was revising gain-on-sale aspects of its securitization accounting. Yesterday, credit card provider Providian Financial filed an 8-K describing forthcoming changes in its past accounting for securitizations due to far more complex issues than Countrywide’s.
It’s away too early to call it a trend. Right now it fits into the category of coincidence, and not a surprising one at that. Call it part of my “increased accounting precision” thesis: if auditors are out there beating the bushes as their newly-invigorated independence permits them, there are going to sloppy practices permitted in the past that are going to be tightened up. Securitizations, an accounting Frankenstein rife with moving parts and estimates, are an area where auditors might not have trouble finding problems.
Back to Providian: several issues in their securitizations. One had to do with their estimated values for interest-only strips, those devilishly volatile leftovers from a securitization that relate to the “gravy” interest exceeding the interest for servicing securitization assets. Providian’s cash flow model for estimating the crypto trust values took into account the level of principal collections expected to occur when forecasting the repayment of the securitized receivables in the Company’s securitization trust. That’s not exactly the right way to model: auditors Ernst & Young advised Providian that it’s better to base them at each valuation date on the contractual rights the related security holders would have for collections of principal if the investor securities were not revolving.
As of December 2001, Providian had revised its modelling methodology to apply a method that got to the same result – except for the effect of applying total payments versus principal-only payments. In January 2002 the Company changed its methodology to partially incorporate the effect of a revolving period into the principal collection estimate – but this constituted a change to a less preferable method of accounting. That would require obtaining a “prefer ability letter” from Ernst & Young and disclosure of the change. Providian didn’t follow through on this in 2002, so it now faces treating the botched change as an error requiring correction by restating the previously issued financials.
Ernst & Young identified found other flawed assumptions used in 2000 and 2001 that deserve error treatment. These include: the effect of the December 2001 triggering of certain spread account funding provisions, which the Company included in its January 2002 cash flow projections, should have been included in December 2001; the modification of cash flow forecasting techniques to eliminate certain averaging conventions for 2000 and 2001; and a change from total payment to principal-only payment assumptions for purposes of forecasting customer loan repayments. (Providian began this methodology in November 2002.)