The Big Lie Lives On

The Big Lie just won’t die. The Big Lie, of course, is The Government Made Me Do It theory of the financial crisis, that the housing bubble whose collapse set off the crisis was the product of government policies encouraging affordable home mortgage credit.

A video emerged recently of presidential candidate Mike Bloomberg espousing the Big Lie, and incredibly, the New York Times is running an op-ed that defends the Big Lie. Most of the op-ed comes verbatim from a new book by Christopher Caldwell. Caldwell has written a remarkably misleading piece about government affordable housing policy. It misrepresents that actual legal requirements, gets the relationship between the GSEs and private securitization market entirely backwards, wrongly implies support from scholarship that is saying something altogether different, and relies on outdated scholarship. I get that Caldwell isn't a housing finance expert, and his book is a trade book on the welfare state, but this is exactly the sort of silliness that happens from drive-by analysis. I'm pretty sure that the Times wouldn't run unsourced climate denial claptrap, but this is the housing finance equivalent. Let me highlight several examples.  

First, Caldwell writes that HUD “required that low-income loans make up 50 percent of the G.S.E.s’ portfolio”.  FALSE.  Caldwell wrongly conflates low-to-moderate income (a group that includes median income borrowers!) with low-income borrowers and ignores that the GSEs met a substantial part of their goals for lower-income consumers by financing rental properties that had nothing to do with the bubble.

To oversimplify a bit, the housing goals require the GSEs to ensure that various percentages of the units whose purchase they finance qualify under three separate goals:  a low-to-moderate income (LTMI) goal, a special affordable goal, and an underserved area goal. A mortgage could count for more than one goal, and the goals were calculated by number of dwelling units, not dollar amount, so a modest property got the same as a fancier one. 

Only one goal, the LTMI goal ever exceeded 50%, but LTMI is not the same as low income. It is everything up to median income for the area adjusted by household size. LTMI does not necessarily include any low income households. Today, for example, in San Francisco it would include a 4-person household with an income of $96,000. Not wealthy, but hardly poor either. Given that the very wealthy do not get mortgages and that the mass affluent are often in the market for mortgages larger than the GSEs are allowed to finance (jumbos), the LTMI market has always been part of the GSE market.  Thus, in 2006, Fannie Mae met a third of its LTMI goal with households at 80%-100% of area median income. 

Now, the special affordable goal is properly characterized low-income, with low-income defined as up to 80% of median income. Today that’s a household earning $77,000 in San Francisco.  The special affordable goal was never set at over a quarter of units financed. Virtually every mortgage that qualified for special affordable also qualified for the LTMI goal, so to the extent that the LTMI goal exceeded the special affordable goal, it was filled almost entirely with mortgages to borrowers earning at least 60% of median. Moreover, Again, on the lower half of the income spectrum, but well above the poverty line. Indeed, it is important to keep in mind that the poor generally don’t get mortgages, as they cannot save for even modest down payments.

As for the underserved areas goal, it is not keyed to borrower income, but to the availability of mortgages in the area in the past. While meant to channel mortgages to capital starved areas, it also financed yuppies’ purchases of properties in gentrifying urban neighborhoods. 

Even more important, perhaps, is that the GSEs, got the same housing goal credit for financing a mortgage for an owner-occupied single-family residence as for each rental unit in building with a multi-family mortgage.  Thus, financing the purchase of a 100 unit low-income rental apartment building would have the same effect as financing the purchase of 100 single family homes. The GSEs were able to cover a large part of their goals for through financing multi-family rental buildings. For example, in 2006, Freddie Mac met over half of its special affordable goal with rental units.  Financing rental properties is an activity with zero connection with the housing bubble. But you wouldn’t know that from the advocates of the Big Lie, who would have you believe that the affordable housing policies were forcing the GSEs to put welfare recipients into McMansions.

Second, Caldwell writes that “The G.S.E.s’ underwriting standards became those of the whole industry.”  WRONG. Caldwell has to make this move in order to explain why the GSEs were somehow responsible for the real carnage, which took place in the private-label securitization-financed market. The problem is that the contagion wasn't from the GSEs to private-label, but the other way around, which is pretty easy to prove. The GSEs lost enormous market share during the bubble years (2003-2006). The market shifted dramatically in those years from the GSEs to unregulated private-label securitization. The GSEs only regained market share in 2007—once home prices were already declining from their spring 2006 peak. If the GSEs were leading the way in dropping underwriting, they would not have lost market share. Instead of leading the race-to-the-bottom in underwriting standards, the GSEs lagged behind and then because of shareholder pressure they tried to play catch-up.

As it happened, the loss rates on GSE mortgages were way lower than for private-label securitizations. The epicenter of the crisis was private-label securitization, and private-label securitization to the extent it had meaningful underwriting standards was not copying the GSEs.   

Third, Caldwell cites scholars in a way to imply that their work supports his interpretation, when it does not.  Caldwell does something very slippery in this piece: he cites the work of some well-respected scholars in a way that implies that they agree with his analytical conclusion. He cites to the work of Atif Mian and Amir Sufi that found that income and mortgage credit growth were negatively correlated. Caldwell presents this citation in the context of discussing GSE underwriting standards, with the implication that Mian and Sufi were talking about the GSEs, when in fact they were talking about the entire housing market, which shifted from the GSEs to private-label securitization in this period. Saying that the crisis was due to an excessive supply of mortgage credit is not at all the same as saying that affordable housing policy was the cause of that excessive supply. But you'd never know that from Caldwell. You'd think that the GSEs were the only game in town, even though private-label securitization was financing a majority of the market by 2006. 

Fourth, Caldwell relies on outdated scholarship that is not supported by more recent literature. Likewise, Caldwell relies on work done by Raghuram Rajan and Viral Acharya et al. that was written in 2010. These are good scholars, but their writings in this area are out of date from a scholarly perspective. Their work was written before the dust had cleared from the crisis and before any of the empirical explorations of the GSEs' role in the housing bubble had emerged. That empirical literature has yet to find any material connection between the GSE housing goals and the bubble. 

And finally there’s this unsourced gem in the Caldwell piece:  “By 2007, high-risk mortgages made up 22 percent of the G.S.E.s’ portfolio, up tenfold from a decade before.”  COME AGAIN? I have no idea what Caldwell means by a “high-risk” mortgage; that is not a standard category of mortgage loans. Perhaps there's some source he's interpreting to mean "high-risk," but I suspect it's based on the presence of a single feature in the loan, without considering the totality of the loan. In any case, looking at 2007 is a totally misleading picture, as the GSEs went chasing shite in 2007 in a way that they had never done previously, and the 1997 baseline ignores the impact of automated underwriting (that had only started to be deployed) that allowed the GSEs to safely take a more nuanced (and less discriminatory) view of risk. 

Let’s be clear: the overwhelming scholarly consensus is that government affordable housing policy had little to do with the financial crisis. Fannie and Freddie certainly made mistakes, but it is simply incorrect to say that they were responsible for the housing bubble. They were followers, not leaders. Yes, the Big Lie still has currency in conservative political circles and their affiliated think tanks (for the Big Lie fits well with their priors that the government shouldn't intervene in markets), but there’s just no serious scholarship that supports it.

There are five fundamental problems with the Big Lie, and they aren’t going to go away from mere repetition of the Big Lie: 

• There is no empirical evidence whatsoever that supports the Big Lie. To the extent that any studies have found connections between affordable housing policies and home prices (and these studies’ findings have been contested), the magnitude of the impact found by the studies has been exceedingly small. But since when has evidence ever mattered when a theory supports someone’s priors?

• The Big Lie cannot explain why the bubble started when it did (2003), as all of the government policies it points to were in place well before then.

• The Big Lie cannot explain why housing finance shifted sharply from GSE securitization to private-label securitization starting in 2003.

• The Big Lie cannot explain the parallel bubble in commercial real estate. Fannie and Freddie finance multi-family residential mortgages.  But there is no government role whatsoever in office buildings, industrial facilities, and shopping malls.  So why was there a bubble in all of these property types, which were financed heavily through private-label securitization?

• The Big Lie cannot explain the parallel housing bubbles that emerged in some, but not all European countries (Denmark, Ireland, Netherlands, Spain, UK) despite incredibly different housing finance systems and different methods for government support of homeownership.

Bottom line: if you’re pushing the Big Lie, you need to have some sort of answer for these five issues. Otherwise, you’re blowin’ hot air, and respectable publications shouldn’t be giving space to folks who do that.

[updated and expanded 2.16.2020] 

Comments

[Barbara Tilney]

However you describe it, the mortgage crisis began when
the borrowing requirements were changed to enable those who could not meet the financial requirements to borrow. Those changes began at the federal legislative level.That some people saw a way to make money from the mortgages that were bundled to absorb shakier mortgages may or may not have been a deliberate unintended consequence but it was a consequence
B

[Adam Levitin]

With due respect Barbara, that's just wrong.

(1) the federal government prescribes almost nothing in the way of mortgage underwriting standards other than for FHA/VA insurance. FHA/VA were a small part of the market and lost market share throughout the bubble. For Fannie/Freddie, the only hard underwriting rules from the federal government are a conforming loan limit (maximum loan size) and an LTV limitation absent private mortgage insurance.

(2) Fannie/Freddie's adoption of automated underwriting in the mid-1qq0s did expand the credit pool, but it did so with a very modest increase in default rates, and that happened years before there was an increase in house prices without an increase in rental prices that can be explained by interest rates--that is before a bubble. While prices marched up from the mid-q0s, the bubble only started in 2003.

(3) Fannie and Freddie were followers, not leaders in the underwriting race-to-the-bottom. They lost market share until 2007, after the bubble had already peaked. If they had led the race, their market share would have grown.

(4) There's just no evidence that suggests that the Affordable Housing Goals were responsible for much of anything, especially as OFHEO didn't enforce the goals. It let the GSEs get away with all kinds of patent gaming, like Freddie Mac renting a portfolio of low income mortgages from WaMu for the time period when its Housing Goal assessment was occurring.

The facts simply do not support the idea that the bubble was driven by any federal legislative action. It was worsened by inaction from federal financial regulators regarding predatory lending, but that's a separate story.