On February 11, Fannie Mae priced its tenth Connecticut Avenue Securities (CAS) risk-sharing transaction. Since the program’s inception in 2013, Fannie has issued $13.4 billion in these notes, covering about $470 billion in newly originated single-family mortgages and obligating the company to pay about $7 billion over the next ten years in premiums and hedging costs (assuming it’s hedging the notes’ short-term interest rate risk, as it should be).

In the initial CAS deals, Fannie took the risk on the first 30 basis points of credit losses (technically “credit events,” but we’ll try to keep this understandable) on a specified pool of mortgages, and issued securities totaling 2.57 percent of this pool that paid investors floating rates of around 3.50 percent over 1-month LIBOR, before hedging costs, for taking 95 percent of the next 270 basis points of losses.

Even these first deals had excessive amounts of coverage and extremely generous pricing to investors, but the terms on the two most recent CAS issues have been so unfavorable to Fannie—committing the company to pay sharply higher interest rates for still greater amounts of unneeded coverage, while requiring it to take more losses itself before the insurance kicks in—as to merit being called giveaways. In the October 2015 CAS transaction, Fannie took the first 50 basis points of loss, while paying investors 1-month LIBOR plus 455 basis points (before hedging costs) to take 95 percent of the next 350 basis points of losses. Then, in the CAS deal done last month, Fannie paid one group of investors LIBOR plus 1175 points (that’s not a misprint) to split the first 100 basis points of losses, and paid two other groups of investors an average of LIBOR plus 500 basis points for coverage against 95 percent of losses over 100 and up to 400 basis points.

The terms and pricing on the recent CAS deals have worsened for three reasons. Most importantly, they are being done not because they make economic sense, but because they’ve been mandated by FHFA, which in turn has been told to do them by Treasury. Second, there is a very limited and highly specialized investor base for CAS transactions (the Urban Institute says 59 percent of the more junior risk-sharing tranche in the October 2015 CAS deal was purchased by hedge funds), so as more deals are pushed on the market the terms for absorbing them deteriorate. And third, the very structure of the CAS securities, by grossly over-insuring the underlying collateral, inadvertently signals to investors that the risk-bearing tranches of these deals contain far more potential for loss than they actually do.

There is more than a little irony in this last point. The standard of protection for CAS transactions—insuring against losses of up to 4 percent of the original balance of the loans in the pool—is derived from losses experienced between 2009 and 2012 on mortgages originated between 2005 and 2008. But these loans, and their losses, come from a highly anomalous period in mortgage lending history, in which control over underwriting and financing shifted to the same Wall Street firms and their supporters, including Treasury, who now insist that Fannie (and Freddie) must insure against an environment that will not, and indeed now cannot, repeat itself.

The huge jumps in the riskiness of mortgages made by all lenders, including Fannie and Freddie, during the 2005-2008 period—and the consequent mammoth losses these loans suffered in subsequent years—were the results of the lethal interaction of two factors: (a) the rise to dominance in 2004 of a financing mechanism, private-label securitization, that placed few limits on mortgage credit risk anywhere along the financing chain, and (b) a deliberate policy choice by banking regulators not to prohibit any of the irresponsible lending practices that took hold in the primary market during this period, including “liar loans,” interest-only ARMs with deep teaser rates, and excessive risk layering. The combination of a very powerful but indiscriminate financing mechanism and no substantive regulation of either the products or the borrowers it served produced four years of undisciplined mortgage lending on a massive scale, culminating in a boom and bust cycle for the mortgage and housing markets whose impact lingers even today.

We clearly learned our lesson from this disaster. The private-label market now is moribund (and unlikely to return as a major financing source for a very long time, if ever), while the Consumer Financial Protection Board has prohibited the types of toxic loans responsible for the large majority of the 2009-2012 credit losses. With neither of their causative factors any longer present, the credit losses on Fannie or Freddie’s 2005-2008 books of business are not relevant benchmarks for the loans the companies are making today.

A much more fitting reference point for gauging the riskiness of these loans is the comparable set of mortgages Fannie purchased or guaranteed from 2000 to 2003 (before lending standards began deteriorating in 2004). The credit quality of the 2000-2003 and 2014-2015 loans is similar, with the more recent arguably being better. The average loan-to-value ratio of Fannie’s 2014-2015 loans is only somewhat higher than its 2000-2003 loans (76% versus 73%), but the average credit score—746 versus 714—is considerably higher. Over 12 years of performance data now exist for the 2000-2003 loans, and they have an ever-to-date default rate of 1.5%, an average loss severity of 34%, and a credit loss rate of 51 basis points.

I was CFO at Fannie during 2000-2003. We were shown risk-sharing deals by Wall Street firms that had much more favorable parameters and pricing than the ones Fannie is doing today, and we did not do any of them (Freddie did a few) because our analysis found their risk-adjusted costs to be too high. We would have rejected the most recent CAS deals out of hand. A simple illustration using the October 2015 CAS transaction—which had better terms than last month’s deal—shows why. Had Fannie insured the same categories of its new 2000-2003 single-family loans as it is insuring now (for the 2000-2003 books, just under half the total) using the structure and pricing of the October 2015 CAS deal, it would have paid at least 7 percent per year (before hedging costs) on an average balance of about $31.5 billion in risk-sharing securities, for total premiums of at least $22 billion for the minimum 10 years the securities would have been outstanding. Based on the performance of all loans in the 2000-2003 books, the CAS-insured loans would have suffered no more than about $5.5 billion in cumulative credit losses, virtually all of which would have fallen under the 50 basis point threshold for risk-sharing, and thus been absorbed by the company. After-tax, the $22 billion in premiums Fannie would have paid to insure against a further 350 basis points in losses (an absurd amount of coverage to begin with) would have cost it over $14 billion in foregone capital, and returned at most a few hundred million dollars in loss reimbursements.

There is no reason to expect the results of Fannie’s more recent CAS transactions on even better books of business to be very different: virtually all of the risks on the loans are likely to end up as losses for the issuer (Fannie), while virtually all of the premiums are likely to end up as income for the investor (principally hedge funds).

Although it is FHFA that has mandated Fannie to “transfer credit risk on at least 90 percent of the unpaid principal balance of newly acquired single-family mortgages in [targeted] loan categories” this year, Treasury is behind this requirement, and it cares not one whit about whether the risk-sharing transactions make economic sense for the company. Treasury’s support for the CAS program stems from its goal of winding down Fannie and Freddie and replacing them with “private market” alternatives. With private-label securitization inoperative, securitized risk sharing is the one remaining tool Treasury has for accomplishing its objective of transferring secondary mortgage risk management (and revenue) from Fannie and Freddie to Wall Street and its customers.

Here, however, is one more heavy irony. Treasury reflexively terms Fannie and Freddie a “failed business model”—even though they dramatically outperformed all other sources of mortgage finance prior to the crisis—yet the credit risk-sharing transactions it is requiring them to do on a massive scale have serious flaws that even the technique’s supporters acknowledge. Those flaws make securitized risk sharing unsuitable for anything other than supplementary credit enhancement when their economics make them competitive.

Opponents of Fannie and Freddie are loath to admit it, but the companies’ method of managing mortgage credit risk is simple, proven, and unequalled. It relies on capital and loss reserves to absorb losses up to some defined level of stress (with that amount now under review in the wake of the financial crisis). Credit risk is diversified by product type, loan characteristic, geography and, importantly, over time. The housing and mortgage markets always have been cyclical: the lifetime loss rates for some origination years are low (or very low), for others they are medium, and for some they are high (or very high). The key to diversification across time is to “bank” the excess returns from the good years, either as loss reserves (to the extent the FASB allows you to do so) or as surplus capital, then draw on that excess capital or reserves to absorb spikes in losses from the bad and very bad years.

This diversification of risk across time is precisely where the risk-sharing model breaks down. Investors are eager to do risk sharing when times are good and risks seem low (although even then the transactions can be prohibitively costly), but risk-sharing investors are less plentiful in times of uncertainty and disappear entirely in times of turbulence. And this fact leads to a second, even more serious, flaw: risk sharing in good times takes what should be future capital out of the mortgage finance system, and gives it to investors who do not re-inject it in bad times, when it is essential and generally either very expensive or not available at all.

The fundamental weaknesses of the risk-sharing model aren’t going to change no matter how many CAS deals Treasury forces Fannie to do. And today’s deals are an extraordinarily wasteful means of making an ideological statement. Fannie is giving away huge chunks of its guaranty fees that otherwise would go to the government as profits if the net worth sweep is upheld, or be returned to the company as capital if the net worth sweep is overturned. Certainly it can’t be Treasury’s objective to transfer revenues from the government to hedge funds, and I hope it is not so cynical as to deliberately be draining Fannie’s revenues, expecting to lose the net worth sweep cases and wanting to make it harder for the company to recapitalize itself.

Irrespective of Treasury’s motivations, however, the CAS transactions give FHFA Director Watt a perfect opportunity to exert his independence as conservator, either by changing their structures so that they no longer dissipate Fannie’s assets or, if that’s not possible (and I suspect it may not be), stopping them altogether.

In remarks to the Bipartisan Policy Center last month, Watt said, “FHFA expects Freddie Mac and Fannie Mae to determine their pricing as though they were holding capital and seeking an appropriate economic return on this capital.” A corollary of this stance is that, as steward of Fannie’s capital, FHFA should not allow Treasury to waste it. With the 50 basis points Fannie currently charges lenders to guarantee their (high quality) single-family loans, it has ample resources to self-insure against unexpected losses, and to devote any excess guaranty fees to the build-up of its capital base if and when the net worth sweep is invalidated. The credit risk management approach Fannie has used for decades—prudent underwriting, and the diversification of risk across product, loan characteristic, geography and time—has proven to work, and FHFA must not permit Treasury to force Fannie to supplant it with a true “failed business model”: programmatic securitized risk sharing.