by Calculated Risk on 3/02/2007 12:51:00 PM

UPDATE: Added Tanta's comments (see bottom).

From the Fed: Agencies Seek Comment on Subprime Mortgage Lending Statement

The federal financial regulatory agencies today issued for comment a proposed Statement on Subprime Mortgage Lending to address certain risks and emerging issues relating to subprime1 mortgage lending practices, specifically, particular adjustable-rate mortgage (ARM) lending products.

The proposal addresses concerns that subprime borrowers may not fully understand the risks and consequences of obtaining these products, and that the products may pose an elevated credit risk to financial institutions. In particular, the proposed guidance focuses on loans that involve repayment terms that exceed the borrower’s ability to service the debt without refinancing or selling the property.

The statement specifies that an institution’s analysis of a borrower’s repayment capacity should include an evaluation of the borrower’s ability to repay the debt by its final maturity at the fully indexed rate, assuming a fully amortizing repayment schedule. The statement also underscores that communications with consumers should provide clear and balanced information about the relative benefits and risks of the products. If adopted, this statement would complement the 2006 Interagency Guidance on Nontraditional Mortgage Product Risks, which did not specifically address the risks of these ARM products.

The agencies request comment on all aspects of the proposed statement and are particularly interested in public comment about whether: 1) these arrangements always present inappropriate risks to institutions and consumers, or the extent to which they can be appropriate under some circumstances; 2) the proposed statement would unduly restrict existing subprime borrowers’ ability to refinance their loans; 3) other forms of credit are available that would not present the risk of payment shock; 4) the principles of the proposed statement should be applied beyond the subprime ARM market; and 5) an institution’s limiting of prepayment penalties to the initial fixed-rate period would assist consumers by providing them sufficient time to assess and act on their mortgage needs.

Comments are due sixty days after publication in the Federal Register, which is expected shortly. The guidance is attached.

From Tanta in the comments:

On page seven, under the definition of “predatory lending,” you find:

“Making mortgage loans based predominantly on the foreclosure or liquidation value of a borrower’s collateral rather than on the borrower’s ability to repay the mortgage according to its terms.”

Now, they’ve been using this kind of language for years; it’s not like this is new to this document. The problem is that, for years, some of us have wanted them to go on to the next logical step, which is to explain just how a loan can be based on the borrower’s ability to repay if it doesn’t include verification of income and assets. What we have found over the years here, you know, is that lenders have come to think that they can just verify the borrower’s willingness to repay, which is what a FICO score and a down payment proxy, but not the ability to repay out of either income or assets. So they keep using that word “ability,” and then they keep going on as follows:

“Risk-layering features in a subprime mortgage loan may significantly increase the risks to both the institution and the borrower. Therefore, an institution should have clear policies governing the use of risk-layering features, such as reduced documentation loans . . . an institution should demonstrate the existence of effective mitigating factors that support the underwriting decision and the borrower’s repayment capacity.”

OK, so which is it? Is the absence of documentation of “ability to repay” mitigated by the appraised value and the FICO score? How is that, exactly, different from making a loan based on collateral value?

“The higher a loan’s risk, either from loan features or borrower characteristics, the more important it is to verify the borrower’s income, assets, and liabilities. When underwriting higher risk loans, stated income and reduced documentation should be accepted only if there are mitigating factors that clearly minimize the need for direct verification of repayment capacity. For many borrowers, institutions should be able to readily document income using recent W-2 statements, pay stubs or tax returns. A higher interest rate is not considered an acceptable mitigating factor.”

I put that last sentence in bold because it is the only outright statement of something you could call an actual rule in this document. But notice how they’re still waffling about the issue of collecting those W-2s, pay stubs, and tax returns. Look, any underwriter worth her salt will tell you that there are, in fact, good credit risk borrowers out there whose income cannot be easily read off one of those documents. The classic self-employed borrower with odd cash-flow patterns comes to mind. But there’s just a world of difference between telling a bank it can, if it finds mitigating factors, override the numbers on the documents, and telling the institution that it can get away with not asking to see them. The whole “stated income” thing gets to be a problem because the industry decided there were only two choices: use the numbers on the tax returns—even if they don’t make much sense as a measure of GMI, and resulted in terrible-looking DTIs—or just let the borrower make things up.

There are two very, very important issues underlying this false dichotomy. One, we’ve seen in action lately, and you can call it the “rep and warranty” problem. If I see your tax returns, but approve you anyway because I did an operating income analysis, looked at the value of your assets, and decided that you should get a loan even though technically your DTI calculates too high, then it is the quality of my underwriting decision that is at stake here. If, however, I tell the LO to throw those tax returns away and resubmit the file as “stated,” then I still get to make the loan, but if it turns out to be a bad idea after all, I can claim to have been defrauded by the borrower. It remains a major mystery why the regulators haven’t caught on to this yet.

Two, there is actually some research out there on the question of “lender-chosen” versus “borrower-chosen” documentation reductions. Moody’s, for instance, has found that programs in which the lender expects documentation from all borrowers, but, after analysis of the application, credit report/FICO, and any other supporting documentation, may waive the income or asset documentation requirement for higher-quality borrowers, perform much better than programs in which the borrower can withhold documentation from the lender and get qualified on a presumption of reduced documentation. Fannie Mae and Freddie Mac, for instance, do this with their automated underwriting systems: the originator enters the application data into the system, which automatically provides itself with a credit report and (usually) data about the property and plausibility of the sales price/appraised value (an internal AVM). It then evaluates the loan, and if it likes what it sees enough, it may report back to the originator that, say, the loan can be counted as “full doc” with just the last pay stub (no W-2s for the last two years), or with just one bank statement showing enough funds to close (instead of three months worth of bank statements showing funds to close plus reserves). The point here is that the borrower doesn’t get to walk in the door and say, “I want one of those loans where I don’t have to give you my W-2 or my bank statements.” That latter thing would be the “borrower-chosen” approach, and those are the ones that perform really poorly. Of course, a “lender-chosen” program will also perform really poorly if the underwriting analysis gives too much weight to DTI or down payment in the initial analysis. (That is, if you’re hanging your hat on DTI, it’s not wise to waive the documentation of the income. If you’re hanging your hat on down payment, it’s not wise to waive the documentation of the source of those funds. You have to do a demonstrated holistic approach in order to do a “lender-chosen” program correctly.)

I honestly can’t see any reason why the regulators wouldn’t at the very minimum require that institutions making reduced doc subprime loans handle this the way the GSEs do, namely only as lender-chosen, not as borrower-chosen, programs which require documentation for most borrowers and waive documents only after a holistic review. But they don’t even raise the issue in the guidance. What’s with that?

Tanta