On Wednesday, Sens. Johnson and Crapo gathered ‘round the GSE-reform campfire and again held hands in hopes of passing housing-finance reform legislation this year. Branding the bill their top priority, the statement came a day after the Wall Street Journal quoted Administration officials from Gene Sperling on down clamoring for quick agreement on a bipartisan bill. The Johnson-Crapo statement combined with a rare White House cry for help convinces me that GSE reform rests on a razor’s edge. The problems are not just political, but also profound policy challenges that can only be resolved with a clear-eyed discussion that balances bipartisan desire for low-cost, fixed-rate mortgages for all with the huge amounts of costly public or private capital it takes to make this dream come too. Fannie and Freddie were and are huge precisely because they were and – absent federal protection – still are highly-leveraged, very risky enterprises. Either the new model is smaller or it’s less risky – you can’t have huge books of business without capital unless you want to do the GSE disaster thing all over again. So, how to have a light lunch and leave room for a bit of cake while crafting the new housing-finance construct?

Johnson and Crapo are, we are told, trying to build the better Corker-Warner mousetrap. That is, they are taking the S. 1217 approach and winnowing it down into one in which a wide array of bond guarantors would replace the GSEs (which would be sent into the sunset in slow-motion liquidation). These guarantors would take a first-loss tranche – Corker-Warner says ten percent – ahead of a U.S. Government (USG) promise to pay investors if the guarantor can’t. But, the financial crisis demonstrated at great cost what happens when the USG provides a backstop – even just an implicit one – without up-front capital buffers from banks and the GSEs. Reflecting this, the legislation has to demand a boat-load of capital ahead of the taxpayer if it wants to honor the “never-again” pledge that trips off the lips of everyone who talks about Fannie Mae and Freddie Mac. How much capital must then there be?

A look at the comparable banking-agency rules shows a boat-load of Titanic proportions. Any bank that takes a first-loss tranche in a structured asset securitization like those proposed in the new GSE model has to hold dollar-for-dollar capital. And, since U.S. banks – especially big ones – are supposed to be not just adequately-, but also well capitalized, the requirement is at least $1.25 for each dollar of first-loss risk exposure. This is totally uneconomic – the first-loss piece has to be priced very, very high for banks to make any money doing this – but this isn’t an accidental technical glitch in the complex capital rules. Bank regulators don’t like first-loss tranches because they’re way risky.

The only way to make first-loss tranches a lot less risky and, thus, perhaps to persuade regulators that those ahead of the USG could hold economic amounts of capital is to make the mortgages in the MBS a lot less risky. The travails of private mortgage insurers during the crisis show clearly just how much risk results from first-loss positions when the loan-to-value (LTV) ratio is over eighty percent. Maybe this risk can go way down if the high-LTV loans are otherwise awesomely clean and the risk would go down still farther if the mortgages collateralizing the tranche are low-LTV ones with similarly squeaky credit-risk profiles. But, the more gilt-edged the mortgage and the lower the first-loss risk, the fewer mortgages make the grade for a USG backstop.

How to square this circle? One could of course go the way of the Hensarling bill, and simply say good riddance to the USG backstop – essentially damn the thirty-year FRM and full-speed ahead. The problem, however, is that full-speed ahead grounds itself on a lot smaller market for the beloved middle-class homeowner. Chairman Hensarling is fine with this, but not so much Sens. Johnson and Crapo and the rest of the Senate Banking Committee.

The other way to solve for capital and credit availability is to reduce the capital requirement, but otherwise ensure that the first-loss holder is a robust bulwark against taxpayer risk. This is easier said than done. It’s the challenge with which Senate Banking has been wrestling as bill sponsors try to duck the question with vague legislative language without sacrificing the taxpayer in ways CBO will score as large upfront budget costs. If lots of bond guarantors are to take the first-loss tranche under all sorts of state and federal rules – if any – the only way around the capital question in this model is to charge a whopping up-front premium for the pleasure afforded the guarantor for its fun with mortgage finance. The higher the premium, the less economic the firstloss piece becomes even if the capital requirement goes down, bringing one back to the same dilemma with which Senate Banking leadership has struggled since at least last summer.

The solution here lies in a proposal unveiled without fanfare last year by the Federal Reserve Bank of New York. It keeps the bond-guarantor model, but assigns this task only to one entity: a new utility expressly chartered to take this on. It would be owned by lenders and capitalized not just by the resources of the utility, but also by recourse to owners – essentially a two-tier federal cushion far more robust than the simple bond guarantor approach, in my view. The utility model also solves for the economic challenge of reconciling capital and shareholder return because investors in a utility – who get service for stock – are content with a lower-than-economic rate of return. Just ask members of the Federal Home Loan Bank System. It’s probably too late for the utility model to surface as a solution to sustaining the thirty year fixed rate mortgage. But, if the legislation is to be not just politically-palatable, but also prudent, Congress has to take the capital challenge on with discipline, not just duck it by deferring to a to-be-determined new regulator.