But here’s the thing: Fannie and Freddie had nothing to do with the explosion of high-risk lending a few years ago, an explosion that dwarfed the S.& L. fiasco. In fact, Fannie and Freddie, after growing rapidly in the 1990s, largely faded from the scene during the height of the housing bubble.

Partly that’s because regulators, responding to accounting scandals at the companies, placed temporary restraints on both Fannie and Freddie that curtailed their lending just as housing prices were really taking off. Also, they didn’t do any subprime lending, because they can’t: the definition of a subprime loan is precisely a loan that doesn’t meet the requirement, imposed by law, that Fannie and Freddie buy only mortgages issued to borrowers who made substantial down payments and carefully documented their income.

So whatever bad incentives the implicit federal guarantee creates have been offset by the fact that Fannie and Freddie were and are tightly regulated with regard to the risks they can take. You could say that the Fannie-Freddie experience shows that regulation works.

Fannie and Freddie had about as much to with the "explosion of high-risk lending" as they could get away with. We are all fortunate that they couldn't get away with all that much of it. It is a fact that their market share dropped like a brick in the early years of this century, except of course for years like 2003, when fixed rates dropped to cyclical lows, refis boomed, and GSE market share shot up again, only to plummet in the years following during the purchase boom.

But they didn't like losing their market share, and they pushed the envelope on credit quality as far as they could inside the constraints of their charter: they got into "near prime" programs (Fannie's "Expanded Approval," Freddie's "A Minus") that, at the bottom tier, were hard to distinguish from regular old "subprime" except--again--that they were overwhelmingly fixed-rate "non-toxic" loan structures. They got into "documentation relief" in a big way through their automated underwriting systems, offering "low doc" loans that had a few key differences from the really wretched "stated" and "NINA" crap of the last several years, but occasionally the line between the two was rather thin. Again, though, whatever they bought in the low-doc world was overwhelmingly fixed rate (or at least longer-term hybrid amortizing ARMs), lower-LTV, and, of course, back in the day, of "conforming" loan balance, which kept the worst of the outright fraudulent loans out of the pile. Lots of people lied about their income (with or without collusion by their lender) in order to borrow $500,000 to buy an overpriced house in a bubble market. They weren't borrowing $500,000 from the GSEs.