In a comment on this post, David Min writes,

So if you want to say that Fannie and Freddie need major reforms because they took on an unacceptably high level of risk, particularly for taxpayer backed entities, welcome to the club. I don’t think there’s anyone who will disagree with you, certainly not anyone I’ve met who’s studied the data even a little bit. On the other hand, if you want to argue that Pinto’s “high risk” loan categories (which you are apparently agreeing are comparable to prime conforming loans) are equivalent to PLS subprime, or even PLS generally, you are completely ignoring all the data.

PLS stands for “private-label security,” the loans originated outside of Freddie Mac and Fannie Mae.

In fact, I probably would not go with Ed Pinto’s definition of high-risk loans. That is because of my bias, based on the pricing models that I helped develop. My bias is to treat the propensity for default as a function of borrower’s equity. I put a lot less weight on factors related to credit history, income, and so forth. Yes, those factors matter at the margin, but you could never convince me to price a loan with a 95 percent loan-to-value ratio as less risky than an owner-occupied purchase loan with an 80 percent loan-to-value ratio, no matter how great the credit score you had on the 95 % LTV loan.

Let me explain here my bias against cash-out refinances. When somebody purchases a home, they have an incentive to negotiate the lowest price that they can get for a home. Therefore, when the appraisal comes in at the sales price, I do not worry about the appraiser’s judgment. Even if the appraiser did nothing to justify the price, the buyer is presumably not an idiot who would pay a big premium over market. But when it’s a cash-out refi, then I am totally relying on the appraiser’s judgment. And I know that the mortgage broker is going to “shop the appraiser” until he gets the “right” number. So I do not trust the appraisal, and therefore I cannot trust the loan-to-value ratio.

A cash-out refi is really a consumer loan, which may or may not have collateral behind it. I do not know how to price consumer loans. I claim to be able to price mortgage loans.

My other bias is against mortgage loans for investment properties. The problem there is you have the “ruthless defaulter.” Somebody whose house declines in value to a point below the outstanding loan balance will often continue to make payments, because they live there. But if you invest in a house and its value falls below the outstanding loan balance, you turn in the keys to the lender.

My idea of a safe loan is a loan-to-value ratio of 80 percent or less that is not a cash-out refi or an investor loan. Give me those characteristics and I am not going to worry about the credit score and the loan documentation. The NINJA loan, where the income, job, and assets of the borrower are not verified, would make me a bit nervous, but the 20 percent or more in equity would probably overcome that. As long as house prices do not fall by more than 20 percent, I don’t expect to lose money on that loan.

On the other hand, if you offer me a loan with a loan-to-value ratio over 90 percent, or a cash-out refi, or an investor loan, I have to price that loan as very risky. All it takes is a slight decline in house prices to make that a bad loan. Any of those types of loans is a risky loan, the way I look at it. (Well, an investor loan with a 50 percent LTV is probably ok).

In that sense, I think all the talk about “sub-prime” is to me a big head fake. When I think of the growth of high-risk loans in the decade leading up to the crash, I think of investor loans (which more than tripled as a percentage of total mortgage loans, according to the work of Avery and others), cash-out refis, and high-LTV loans.

In my opinion, the industry became way too obsessed with FICO scores. I was a big fan of credit scoring, but mostly because for low-risk loans I wanted to bypass the human underwriters, who I suspected added little value.

I will bet that low-risk loans, by my definition, had a default rate of much less than 5 percent, even if they were originated in 2006 or 2007. I may be overly optimistic about that, but I would like to see the numbers. My sense is there were not all that many loans done at that time that met my criteria for low risk.

Think of a Venn Diagram. Circle A includes loans that are high-risk by my definition. Circle B includes loans that are risky by other criteria, such as pay-option ARMs NINJA loans, or loans to subprime borrowers. My guess is that there was huge overlap between circle A and circle B. However, if you gave me a choice, I would prefer a portfolio of loans outside of circle A, even if they include some loans in circle B. I think that most other people, when they talk about risky mortgage loans, make it sound as if they would be fine with loans outside of circle B, even if they were in circle A. I think that’s wrong. But you have to slice and dice the data more carefully than I have seen in order to know for certain.