This exchange on twitter caught my eye.
There’s nothing particularly outrageous here. A generous interpretation of the exchange includes no technically false statements. Yet, I think, this issue with Texas and home equity borrowing is a classic case of how our existing paradigm affects our acceptance of evidence, and how the flawed paradigm of the conventional wisdom on the housing bubble leads to wrong policy choices supported with confidence.
This all starts with a Dallas Fed paper that quantifies the effect of a rule unique to Texas that limits cash-out refinancing to 80% of the home value. They compared homes near the border in Texas and in the neighboring states and provide solid evidence that the rule was associated with default rates in Texas a few percentage points lower than in the neighboring states. All good and well.
As far as I can tell, they make no strong statements of any kind in the paper about the effect on home prices. In a summary article of the paper, they conclude:
The data suggest that the tighter regulations in Texas helped keep underwater mortgages and default rates from rising by as much as they did elsewhere. By extension, lower default rates and fewer underwater homeowners might also have helped Texas avoid the subsequent sharp drop in home prices other states experienced.
To be sure, these benefits did not come without attendant costs. Just as the restrictions helped Texas navigate the housing downturn, the same restrictions could have constrained consumer spending growth during the boom by preventing homeowners from fully tapping their housing wealth. At the same time, this may have helped limit swings in consumer spending. Moreover, the inability to access housing wealth may have driven some credit-constrained Texans to more-expensive credit card debt, unsecured consumer debt or even payday loans. Any estimate of net benefit of Texas’ home equity regulations must also account for such costs.
Notice the second sentence there. They have no evidence of the effect on prices. It’s a leading speculative suggestion. Perfectly reasonable in isolation. I would have been comfortable making that kind of statement about the paper if I had written the summary. And, it is followed by a commendable set of caveats, which reflect an important wisdom that comes from orthodox finance - illiquidity sucks and it has costs. It is a wisdom frequently forgotten in housing and mortgage lending policy discourse.
The results made their way into the Washington Post in an article titled, “How Texas escaped the real estate crisis”, which dropped the caveats and set the modest Texas default findings inside the familiar morality tale of profligate Americans spending fake wealth. “Mimicking Texas would be the perfect opportunity to get our home-equity debt addiction under control and learn to live as an 80 percent nation.”
Fast-forward to the present, where, again, Bill McBride doesn’t actually say, “limiting cash out refinancing prevents price volatility”, but instead, Rick Palacios mentions declining prices and Bill replies (paraphrasing), “Say, that reminds me about this cash out refinancing thing.”
The thing is, the cash-out rule probably does stabilize Texas default rates a bit, but if anything, it is evidence that default rates weren’t causal to collapsing prices. Here are price changes in Texas compared to its neighbors, with Florida added for scale.
In the left panel, relative gains from 2002 are shown. Texas prices moved along with Oklahoma as the most moderate of the states shown. Could the 80% rule have been a factor there? Sure. There is some plausibility that illiquidity lowers prices. This is standard finance.
However, looking at prices indexed to their peaks, in the right panel, Texas is pretty typical compared to its peers. There is no evidence here that the 80% rule reduced the decline of prices in Texas.
In fact, there is little difference between Texas and its neighbors. One could try to engage in a series of controls, to try to quantify the connection between prices and the 80% rule. I’m not going to do that because it is clear that, relative to the scale of rising and declining prices that characterized the housing boom and bust in places like Florida, it is insignificant. It isn’t worth quantifying. There were fewer defaults, and fewer defaults had little or no effect on home prices in Texas during the bust.
A majority of the country from 2000-2012 basically looked like Texas, and then there were a number of outlier markets where prices fluctuated wildly. Frankly, the idea that the Texas housing market was different than Florida, or Arizona, or California, because of the 80% rule isn’t just a bit of overstatement. It’s ridiculous.
Some Closing Thoughts
The rhetorical process here has layers of irony. One of the points that my housing project has driven home to me is that our sense of what’s true, of what’s evidence, etc. can be highly sensitive to our priors.
It affects the questions you ask and the focus of your attention. It affects the depth of confirmation required to accept something as empirical fact. Compare the introduction of the Dallas Fed paper (“Given the role of excessive borrowing in precipitating the housing crisis, economists and policymakers have focused significant attention on effective regulations to curb unaffordable mortgage debt. But evidence whether such regulations indeed work remains thin.”) to the introduction to another Fed paper on housing supply:
“While it is now clear that too much housing was built in the US in the boom phase, identifying how much and where overbuilding occurred remain important issues.” The statement is uncited. As far as I can tell, it is uncited because nobody had actually published on the question. They were all too busy trying to quantify the effects of lending on home prices. On the demand side, an infinite number of monkeys were typing in search of evidence while on the supply side, the conclusion (becoming the presumption for subsequent research) could be reached through osmosis.
So, the Washington Post article on the 80% rule, which began, “It's one of the great mysteries of the mortgage crisis: Why did Texas -- Texas, of all places! -- escape the real estate bust?” dripped with an implied paradigm, which is sometimes, but not always, supported by strong evidence: There were forms of reckless borrowing. That borrowing fueled a bubble. Where bubbles developed (with “phanton wealth” that funded excessive consumption), they had to reverse. Reckless borrowers who had to sell because they had overcommitted were a source of downward price pressure. That was exacerbated by high LTV borrowing. So, reckless borrowing causes home price crashes. And, of course, it makes perfect sense that Texas avoided the boom and bust with this “one weird trick”.
The acceptance of that evidence is entirely the result of the premise. But, then it becomes evidence in defense of the premise, even though there literally is no evidence of it.
To further the irony, much of the research that establishes a causal role for mortgage activity in the housing boom requires controlling for regional differences. In other words, erasing the difference between Florida and the other states in Figure 1. To be generous, this is because to make a careful assessment of the affects of mortgage access, one must look past regional differences that could have a number of influences, to carefully quantify the true effects of credit access.
So, what the research does is literally erase 80% of the variance in home prices, and then find the most important sources of the remaining 20%, of which lending rules are found to have a significant effect. But, again, the carefulness of the methods gets lost when those methods are cited rhetorically, and through the sheer repetition of implied connections, mortgage lending sidles into that complex conventional story of the boom and bust as the reason Florida differed from Texas. And, then the non-evidence of the effect of the 80% rule sidles in next to it so that “the one weird trick that Texas used to avoid a housing boom and bust” is non-evidence applied to erased evidence.
Furthering the irony, Texas didn’t avoid the housing bust. The severe tightening of credit scarred working class home values across the country regardless of what had happened in the boom. Here are relative home prices of lowest price quintile of ZIP codes and the highest price quintile of ZIP codes in Dallas and Oklahoma City (to continue the comparison to neighboring states).
On average (especially in value-weighted indexes), home values in Oklahoma and Texas didn’t decline much during the bust. But the working class home owners, who Texas so valiantly protected from liquid access to their wealth, were decimated just like everywhere else. By the end of 2013, high end OKC and Dallas were level with values at the peak of the market. The low end had suffered losses of about 15%.
So the non-evidence about the 80% rule is applied to the erased evidence about the difference between Texas and Florida, to explain the non-existence of a major, massive thing that actually happened which we have communally chosen to not notice.
And the ironic cherry on top is that what destroyed working class home values - what created the major, massive catastrophe which we chose not to notice - was the harsh imposition of the “one weird trick” of tight mortgage regulation that destroyed owner-occupier buyer demand in working class housing markets.
I’m sorry. My plan was to take a civil tone on this post, but if you’ll excuse me, I’m gonna go put on my “Joker” makeup.
All the freaking unlearning
This is what makes my work so complicated. I have a very simple story to tell. A lack of housing supply makes prices higher. A lack of mortgage access makes prices lower temporarily. But it also makes construction lower, and thus, over time, rents and prices end up higher again. Empty your mind of presumptions about the housing market, and this is a basic financial framework that anyone can understand about supplying a marketplace.
The challenge is the “emptying your mind” part. There are a million things like Texas’ “one weird trick” that block an understanding of the American housing market, and the more you know, the harder the process of unlearning is.
Very simple and effective arguments!