Freddie, Fannie Take on Big Loans, Risk

"Given the lack of historical precedent for current conditions and the fact that models are estimated based on historical experience, enterprise models have become less reliable and require greater management judgment, increasing the potential for error in pricing and other metrics," the OFHEO said in last week's annual report to Congress.

And then there are the specific problems of measuring jumbo-loan risk, which is also hard to quantify because the GSEs have never dealt with it before.

So basically, in a very risky environment they're taking on an unknown -- but likely high -- amount of extra risk.

In addition, the prices of houses that require jumbo loans are likely to keep right on falling: If the run-up in housing prices was purely a bubble, the national market will still fall another 30%. But the prices won't drop evenly, but instead will be largest in the areas with the biggest run-ups in housing prices.

Those are precisely the areas where the GSEs will increase their loan limit: Over 50% of the loans will be in California, according to OFHEO.

If owners of multimillion-dollar properties can't handle their mortgages and decide to walk away, taxpayers have a problem.

The GSEs and federal home-loan banks hold over $1 trillion more debt than the federal government's publicly held debt.

The Fed wouldn't allow the failure of Bear StearnsBSC; there's no way it could allow the failure of these goliaths, either. So they have little incentive not to take on increased risk: If, by some miracle, the jumbo loans prove profitable, shareholders -- and homeowners with six-figure salaries -- gain. Should jumbo loans serve only to temporarily prop up an already bursting bubble, taxpayers lose.

The possibility of a government bailout puts the risk squarely on our shoulders. And with risk management mostly in private hands -- the legislation was pushed through before the OFHEO could be given more meaningful regulatory powers -- it may be a heavy burden.