Bond salesmen, peddling trillions of dollars of Fannie Mae and Freddie Mac fixed-income securities to investors all over the world, said something like this: "You can't go wrong buying this, because it is really a U.S. government credit, but it pays you a higher yield! So you get more profit with no credit risk."

What the bond salesmen told the global investors was true, as events have fully confirmed.

The Treasury Department's financial support of Fannie and Freddie is unlimited. The largest owner of their obligations is now the Federal Reserve.

Paying the foreign and domestic creditors of Fannie and Freddie at par has an estimated cost to American taxpayers in the range of $200 billion to $400 billion. It will be much more than the $150 billion cost of the notorious savings and loan collapse of the 1980s. Both collapses, it should be noted, reflect the yearning of the government to sponsor loans on housing.

The giant GSE risk turkey has been roosting in the dome of the U.S. Capitol ever since Fannie and Freddie entered regulatory conservatorship almost two years ago. It won't go away, so the elected representatives of the people cannot forget the mistakes they made in fattening it up so much.

Nonetheless, the majority of the representatives just passed a 2,300-page financial "reform" act that made sure to avoid addressing the fundamental problem of Fannie and Freddie.

The taxpayer bailout of Fannie and Freddie is without question a government intervention to save previous government interventions.

Since the entire intellectual premise of the Dodd-Frank regulatory expansion act is that government intervention in financial markets will improve things, it was obviously difficult for its supporters to admit that the massive intervention represented by Fannie and Freddie turned out to be a massive government blunder.

Does the 21st century need GSEs in order to have securitization of prime mortgage loans? No, it doesn't.

The future mortgage finance system should have a robust private secondary market for the largest segment of the business: prime, middle-class mortgage loans. In this market, private investors should put private capital at risk, and prosper or lose as the case may be. This is the most obvious case where the risks are manageable and no taxpayer subsidies or taxpayer risk exposures are required or desirable.

There may, decades ago, have been a case for GSEs to guarantee the credit risk of prime mortgage loans in order to overcome the geographical barriers to mortgage funding, which were themselves created by government regulation. There was lately a case for using the GSEs to get through the financial crisis which they themselves did so much to exacerbate. But as we move into the future mortgage finance system, the prime mortgage market should stand on its own.

A private secondary market for prime mortgages should have been a natural market development. Why did it never develop? The answer is obvious: no private entity could compete, or can now compete, with the government-granted advantages, and now the huge explicit subsidies, of the GSEs. There can be no evolution of a private prime mortgage loan market while the GSEs make private competition impossible.

The core issue about GSEs is this: You can be a private company, with market discipline; or you can be part of the government, with government discipline. But you can't be both.

As recommended by the Congressional Budget Office, Fannie and Freddie should now be on the federal budget. Honest, on-budget accounting would give Congress a strong incentive to junk the failed GSE model and proceed to restructure Fannie and Freddie on the correct principle of "one or the other, but not both."

Julius Caesar famously wrote that Gaul overall was divided into three parts ("Gallia est omnis divisa in partes tres"). The fundamental reform of Fannie and Freddie should be the "Julius Caesar strategy": divide them into three parts.

The first part, unfortunately, must be two liquidating trusts or "bad banks" that will bear Fannie and Freddie's deadweight losses, the nearly $150 billion already spent by the Treasury and the additional embedded losses that will be recognized over time, unjustly but at this point unavoidably borne by the taxpayers.

These trusts can be modeled legally on the structure used for the privatization of Sallie Mae in 1996.

The second part of Fannie and Freddie should be formed by the privatization of their prime mortgage loan securitization and investing businesses.

All of their intellectual property, systems, human capital and business relationships should be put into truly private corporations, sold to private investors, and sent out into the world to compete like anybody else--sink or swim, flourish or fail.

They will be free to do anything they think will create a successful business--except trade on the taxpayers' credit card.

The final element of the former Fannie and Freddie consists of those activities that are properly those of the government: such as providing housing subsidies in one form or another and providing nonmarket financing of risky loans.

These should be put explicitly in the government. They should be fully subject to the discipline of congressional approval.

Funding for these activities would have to be appropriated by Congress in a transparent, democratic way.

These governmental functions of Fannie and Freddie should be merged into the structure of the Department of Housing and Urban Development along with the government mortgage programs of the FHA and Ginnie Mae.

In sum, the GSEs should be divided into three parts: liquidating trusts, private mortgage businesses and a government agency inside HUD. No GSEs would be left.

Alex J. Pollock is a resident fellow at AEI.