It should sound familiar: in a severe housing market collapse—with home prices falling by 25 percent—Fannie Mae and Freddie Mac are able to remain profitable on an operating basis, with their revenues exceeding their expenses. But temporary or estimated non-cash accounting entries booked by their conservator, FHFA, cause them to need to take over $100 billion in senior preferred stock from Treasury to maintain a positive net worth. The engineered “bailout” leads long-standing critics of the companies to label them a “failed business model,” and insist they be replaced with some unspecified alternative secondary mortgage market financing mechanism that allegedly will be better and more reliable (but in reality will not be).

This first happened in 2008-2011, after Treasury forced Fannie and Freddie into conservatorship and they came under the control of FHFA. And it happened again last week, when FHFA released the results of the Dodd-Frank “Severely Adverse scenario” stress tests done on the companies’ December 2015 books of business.

The second episode followed a predictable script. Bloomberg quickly put out a headline, “Fannie, Freddie Could Need as Much as $126 Billion in Crisis.” Its article began, “Fannie Mae and Freddie Mac could need as much as $125.8 billion in bailout money from taxpayers in a severe economic downturn, according to stress test results released Monday by their regulator. The Federal Housing Finance Agency said that the government-controlled companies, which back nearly half of new mortgages, would need at least $49.2 billion.” Senator Bob Corker (R-TN)—a long-time opponent of Fannie and Freddie—came forth on cue with his analysis and recommendation the next day. “This stress test is another reminder of our housing finance system’s failed model: taxpayers are on the hook during bad times, but investors benefit greatly during good times,” he said. “Reforming Fannie and Freddie remains the last major piece of unfinished business of the financial crisis, and this news highlights the need for comprehensive housing finance reform.”

Except it wasn’t, it didn’t, and Bloomberg got both the headline and the lede wrong. The headline should have read, “Fannie, Freddie Pass Stress Test; FHFA Fails It.” And the lede should have been, “Stress tests conducted on Fannie Mae and Freddie Mac this year by the Federal Housing Finance Agency showed sharp improvement from both 2014 and 2015, with projected credit losses now less than the companies’ revenues. Fannie Mae and Freddie Mac’s 2016 stress test results also were markedly superior to the results of 2016 stress tests conducted by the Federal Reserve on the nation’s 33 largest banks.”

As with so many other issues related to Fannie and Freddie, the facts about the Dodd-Frank Severely Adverse stress test—both for the companies and the banks subject to the test—are “hiding in plain sight,” readily available to anyone wishing to form their own independent opinion about them.

The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 requires the Federal Reserve to conduct annual stress tests of bank holding companies with assets of $50 billion or more. In these tests, the Fed projects six categories of income and expense (and the resulting total net income or loss) over a nine-quarter period, based on three different scenarios—baseline, adverse and severely adverse—developed annually by the Fed. The Fed has conducted Dodd-Frank stress tests on banks since 2013. FHFA implemented Dodd-Frank’s stress test requirement for Fannie and Freddie, using the Fed’s scenarios, in November of 2013, publishing the first results of those tests in April 2014.

For the 2016 stress tests for banks, Fannie and Freddie—conducted using data as of December 31, 2015—the Fed’s severely adverse scenario has real GDP declining by 6.5 percent through the first quarter of 2017, unemployment rising to 10 percent by the middle of 2017, and home prices falling by 25 percent through the third quarter of 2018. The Fed released the results of the 2016 bank stress tests on June 23; FHFA released the 2016 Fannie-Freddie stress test results on August 8.

The most striking results from the two sets of stress tests were the magnitude and trends of their respective credit losses. The 33 banks with assets of over $50 billion had $385 billion in 2016 stress credit losses—6.1 percent of their average loan balances and nearly half again as much as the $267 billion they were projected to have in 2014. In contrast, Fannie and Freddie had just $27 billion in 2016 stress credit losses—0.5 percent of their average loan (mortgage) balances and barely one-quarter of their $92 billion in stress credit losses projected in 2014.

Fannie and Freddie’s 2016 stress loan losses, in fact, were $3 billion less than their $30 billion in projected net revenues, meaning that on an operating basis the companies remained profitable during the stress period, just as they had throughout the financial crisis. Objectively, Fannie and Freddie “passed” their stress test. Yet as had been the case during the crisis, that was not the result FHFA (or Treasury) wanted to portray. And FHFA knew that by using the same types of non-cash losses it had relied upon to force the companies to draw $187 billion from Treasury during 2008-2011—this time helped by the companies’ near-total absence of initial capital because of the net worth sweep—it could produce the media headlines and public reaction it, Treasury, and Fannie and Freddie’s opponents were looking for.

Unsurprisingly, FHFA began with the loss reserve. With only $27 billion in stress credit losses, Fannie and Freddie’s loss reserves still rose by a combined $39 billion over the stress period. We can compare this with the reserve increase for the 33 banks subjected to similar tests. In the four Dodd-Frank bank stress tests done to date, the average credit loss was $327 billion, and the average increase in bank loss reserves was $23 billion. Fannie and Freddie’s 2016 credit losses were less than one-tenth the four–year average of the banks, yet the addition to the companies’ loss reserve, in dollars, was larger by more than half. Had the same percentage of Fannie and Freddie’s credit losses been added to their loss reserves as was added to the banks’, the companies’ reserves would have increased by $2 billion, not $39 billion.

Next, the $11 billion in non-cash market value losses shown for Fannie and Freddie during their 2016 stress test amounted to 1.6 percent of their $700 billion combined mortgage portfolios; bank market value write-downs of $29 billion in the 2016 test, on their $6.3 trillion loan balance, were one-third that magnitude, at 0.5 percent.

The Dodd-Frank stress test includes two categories of loss that the Fed applies to fewer than one-quarter of the banks it tests. The six bank holding companies with “large trading and private equity exposures” (JP Morgan Chase, Bank of America, Citigroup, Wells Fargo, Goldman Sachs and Morgan Stanley) are subjected to a global market shock, while these six institutions plus State Street Bank and Bank of New York Mellon, with “substantial trading, processing or custodial operations,” also have a counterparty-default scenario component. Like the other 25 banks with $50 billion or more in assets, Fannie and Freddie fall into neither of these categories, but FHFA nonetheless used both of them to add a further $25 billion in estimated non-cash losses to their 2016 stress results. FHFA did not disclose how it arrived at that amount.

Finally, there was the deferred tax asset reserve. This category appears nowhere in the Fed’s stress tests, but FHFA trotted it out—despite the companies’ profitability on an operating basis during the stress period, and their indisputably becoming more profitable after the hypothetical stress is over—to suggest the possibility of another $77 billion in non-cash losses “depending on the treatment of deferred tax assets”. FHFA gave no rationale for the creation of such a reserve (it has none), but by including it in the tabulation of stress test results FHFA was able to publish a high-end “bailout” figure of $126 billion, which the media dutifully reported.

Any serious evaluation of Fannie and Freddie’s 2016 stress test, however, must differentiate between non-cash entries and credit losses. For as long as they remain in conservatorship, the only aspect of the stress test the companies can control is the quality of the loans they own or guarantee at the time the test begins. Here they have made remarkable progress. As of the end of 2015, 84 percent of Fannie’s $2.8 trillion in single-family loans had been acquired after 2008—comparable data aren’t available from Freddie—and the credit quality of those loans has been superb (perhaps even too good from an affordable housing standpoint). The Fed publishes stress losses by category, so we can compare banks’ residential mortgage losses with those of Fannie and Freddie. For 2016, the $1.2 trillion in first-lien residential mortgages held by the 33 largest banks (which account for 80 percent of all banking assets) produced stress losses of $38 billion. Subjected to the same test this year, Fannie and Freddie’s combined $5.2 trillion in single-family loans—over four times the size of the large banks’ holdings—had much lower stress losses, at $27 billion.

The real message of FHFA’s 2016 Dodd-Frank stress test, then, is that Fannie and Freddie have re-emerged as reliable, efficient and low-risk credit guarantors. FHFA is being a good soldier for Treasury in trying to disguise this fact by smothering the companies with non-cash expenses of $75 to $152 billion (depending on whether the deferred tax asset reserve is included) in the stress test. And while that may fool the media and the casual observer, FHFA is fighting a losing battle. Fannie and Freddie only will get stronger with time, and FHFA will run out of ways to hide it.

FHFA is the one failing the stress test. It would not be at all difficult for FHFA to do a true stress test for the companies—one that does not require (or allow) estimates or assumptions of future losses to determine its results. FHFA could subject Fannie and Freddie’s books of business to a 25 percent fall in home prices over whatever time period it chooses, then project their income and expenses, by major category, for as many quarters as it takes to produce a peak dollar amount of losses. The exercise would be all cash flows, without estimates or timing differences, and the maximum loss amount reached would be the basis for setting the companies’ capital requirement.

Yet FHFA is highly unlikely to devise and conduct such a test as long as it is under the influence and control of Treasury, because Treasury does not want it to. A no-gimmick stress test would make it readily apparent that—contrary to Treasury’s fabricated claim that Fannie and Freddie have “fatal structural flaws” requiring them to be wound down and replaced—the companies could in fact be very effective credit guarantors, posing little or no risk to the government, with far less capital than Treasury, Senator Corker and others insist they hold. Derived from an honest, FHFA-run stress test, credible capital requirements for Fannie and Freddie would make it difficult if not impossible to stampede Congress into replacing them with alternatives that serve a narrower range of homebuyers at higher costs than Fannie and Freddie would be able to do. The companies’ opponents know that, and their plan is to stick with fiction over fact for as long as they can. So far, it’s working.

Two things can change this. One is a ruling or rulings against Treasury and FHFA in the net worth sweep cases, which would force FHFA to conclude that it must follow the law and start complying with its statutory duties as conservator of Fannie and Freddie, independent of Treasury. (Setting capital standards for the companies, using a true stress test, would be a top priority of an independent FHFA.) The other is for mortgage reformers and policymakers to open their eyes to the deceptions being carried out in front of them. Whether it’s the 2008 takeovers of Fannie and Freddie disguised as a rescue, the artificial expenses booked by FHFA during 2008-2011 that forced the companies to take $187 billion in senior preferred stock they didn’t need and aren’t allowed to repay, or the phony FHFA Fannie-Freddie “stress test” being peddled to the public in the guise of an honest one, the curtain has been pulled back on the activities of Fannie and Freddie’s opponents. Anyone can now see behind it—provided they wish to.