While I await the release later today of the Administration’s white paper, I have received from Princeton University Press an advance copy of Guaranteed to Fail, a new book by Viral Acharya, Matthew Richardson, Stijn Van Nieuwerbergurgh, and Lawrence J. White. I will excerpt more below, but here is a quick summary.

1. Freddie Mac and Fannie Mae (also known as the GSEs) were bad. Bad, bad, bad, bad, bad.

2. Still, the mortgage market needs some sort of government guarantee to be efficient.

3. The government guarantee should be restricted to standard mortgages for purchasing homes, rather than to facilitate speculation in housing. No sub-prime loans. No cash-out refis. No high loan-to-value ratio loans. No second mortgages. No loans for non-owner-occupied homes.

4. We need a totally new GSE public-private hybrid to implement this policy.

Over 3/4 of the book is about point 1. Since I mostly agree, I was skimming.

I would not have agreed with (2) before reading the book, and they failed to make a case that was compelling enough to change my mind.

However, if we do have a government guarantee going forward, then I think that (3) is the critical point. I endorse that view without any reservations.

The question is whether the government can remain committed firmly to (3). If it can, then I think we should just keep Freddie and Fannie as they used to be, before they loaded up on sub-prime. That would be more efficient than starting up a new agency from scratch.

And if government cannot remain committed to (3), so that somewhere down the road Congress is going to go loony over “affordable housing goals,” then the proposed new-fangled private-public hybrid will turn out to be bad. Bad, bad, bad, bad, bad.Now for the excerpts.
On p. 61, they quote from a CBO study of the costs and benefits of Freddie Mac and Fannie Mae, written in 1996:

Once one agrees to share a canoe with a bear, it is hard to get him out without obtaining his agreement or getting wet.

I was just amused to see that sort of writing from a bureaucracy.

On pages 80 and 81, they tell the story for how the GSEs came to be bad, bad, bad, bad, bad. Discussing the combined holdings of Freddie and Fannie, they write

in 1997…Of the $481 billion portfolio, less than $10 billion was in private-label securities backed by the riskier nonprime mortgages…

Move forward 10 years to year-end 2007: Fannie and Freddie had essentially tripled in size…But the hidden fact underlying this growth…was…Fannie’s and Freddie’s foray into riskier mortgage portfolios was now $331.7 billion, or 22% compared to just 2% 10 years earlier.

Anyway, when I was at Freddie in the late 1980’s and early 1990’s, I did not think we were bad, bad, bad, bad bad. I thought we were doing a decent job of protecting the interests of taxpayers as well as shareholders. Given the way that the GSEs were conducting business at the time, my perception was reasonable. I left for idiosyncratic reasons having nothing to do with corporate strategy. Subsequently, the GSEs went all in for junk mortgages. Even that might have been ok if they had set aside an appropriate level of loss reserves and increased their capital buffer. But they left that responsibility to the taxpayers.

On p. 143, the authors make the case for a government guarantee.

while some parts of the corporate debt market could falter, there is usually an alternative source of financing for corporations…There is no plan B for the mortgage market.

Hello? Do banks not have deposits? Anyway, they say

To the extent that guarantees enhance liquidity of the secondary mortgage market, the cost of financing mortgages will be lower.

No doubt. Everyone who borrows money with a government guarantee behind it is going to enjoy a lower interest rate. But the authors never say why taxpayers should be privileging mortgage debt over other uses of capital. The “unseen” of cheap mortgage credit is the higher cost of funding new businesses or capital expansion of existing businesses.

On p. 153-154, they give us their key reform proposal.

there is a viable private option, in which the GSE’s disappear, but all conforming mortgage-backed securities would still be guaranteed. The idea would be to form a public-private partnership for the guarantee business…

the securitizer would purchase a fraction of its insurance from a monoline insurance company and the rest from a newly formed government entity, which we will call the Government Mortgage Risk Insurance Corporation (GMRIC). As with terrorism risk insurance, the private insurance market would help to establish a market price for mortgage default risk. The newly formed GMRIC would charge a fee based on this market price. This would ensure that the government receives adequate compensation for the credit risk–a key difference with the precrisis approach.

The first thing that occurs to me about GMRIC is that it rhymes with gimmick. The second thing that occurs to me is that the day that Gimmick opens for business, the goal of the private sector will be to figure out how to stick Gimmick with more risk than it realizes it is taking. As the authors themselves write elsewhere (I cannot find the passage), like water flowing down hill, risk finds its way to where there is a government guarantee.

The only fundamental difference between Gimmick and the existing GSEs is that Gimmick would not be allowed to hold loans in its portfolio. It would take only credit risk, not interest-rate risk. I personally would rather see interest-rate risk managed and regulated at Freddie Mac and Fannie Mae.* That risk has to go somewhere, and there is a good chance that it will wind up at the least-carefully regulated Too-Big-to-Fail bank out there.

(*Note, however, that my preferred scenario would be no government involvement, in which case I suspect that the 30-year fixed-rate mortgage would become more expensive, so that borrowers would choose to share a bit more in interest-rate risk, with, say, loans with rates than adjust every five years.)