Thursday, August 31, 2017

Housing: Part 253 - Great new paper on subprime lending

Scott Sumner found a great new paper that looks at lending during the housing boom.

I have been working with various sources of information that demonstrate how there was no marginal increase in homeownership during the private securitization boom, rates of first time homebuyers were actually declining in 2004-2007, and mortgages throughout the boom were going to young professionals with college degrees and high incomes.

Frankly, this information isn't a secret.  It's clear in basic Census Bureau data, Survey of Consumer Finances, etc.  But, since a flawed premise regarding the housing bust was broadly accepted early, all of this data is generally simply ignored.  It's kind of absurd that I have any business being associated with its discovery.  It's unfortunate that it has been left to me to complete this new narrative of the boom and bust, and I do not have the full set of statistical skills and knowledge to make the best case for some of this work.  I am grateful that others have been filling in these gaps in knowledge.

In any case, this paper, from Stefania Albanesi, Giacomo De Giorgi, and Jaromir Nosal, digs into mortgage data with various methods to control for the effect of age among homebuyers, and finds some direct corroboration for these other sources.  There was no shift in risk-taking to homeowners among mortgage lenders.  And, in fact, they find that much of the rise in delinquencies was among financially secure investors.

Now, the consensus has coalesced around the idea that everything was reckless.  So, rising homeownership and expansion of lending to financially insecure buyers can be the reckless cause of the bust.  Or, falling homeownership and expansion of lending to financially secure investors can be the reckless cause of the bust.  Heck, in the decade's worth of a housing depression we have imposed on ourselves since then, many people have convinced themselves that there is a new housing bubble fed by institutions paying all cash for existing homes.  At this point, I'm convinced that if there was a surge of new housing units being built, by hand, by the owners cutting down the trees on their own land with hand saws, and thatching their own roofs, the Wall Street Journal and the Financial Times would post articles about how this calls for tighter monetary policy.

This paper provides support that the crisis had little to do with reckless lending, but detractors can still argue that investors tend to default more in the face of collapsing prices, so that even that lending was reckless.  That's all well and good.  We can have that debate.  But, we need to be clear that this was not the story that filled newspapers in 2007 and 2008 when everyone was standing blithely aside as home prices dropped by 1% per month, for months on end.  The story that led us to accept a crisis was that millions of buyers never had a chance at making their mortgage payments, and that was the cause of the inevitable collapse.

What really happened was that new homeowners were already in decline, long before we got serious about imposing a collapse on ourselves.  We determined that speculators and lenders needed some discipline.  There were a large number of recent young new owners and investor buyers who tend to be more prone to default when prices collapse, and we engineered that collapse, all the while complaining that they did this to us.  And, as time passes, the idea that lenders were throwing caution to the wind seems to be losing to the evidence.

To be fair, there is clearly much evidence that in 2006 and 2007, in some respects, there were sharp shifts in underwriting.  But, if you look closely at many of these complaints, they are complaints about a lack of documentation or complaints about investor buyers engaging in various sorts of misstatements.  This was a strange period, though.  The number of buyers was collapsing, which is a strange thing to see if underwriters are being extremely lenient.  I think, oddly, what we were really seeing was an exodus of previous owners out of the market, and the buyers that remained tended to be more leveraged and more likely to be investors.  This should have been obvious, since, by the time prices were collapsing, housing starts were already at recession levels.  The reason prices collapsed after mid-2007 is because the housing market had already absorbed as much decline as it could without breaking.

Selections from the paper:

Our analysis also reconciles the pattern of borrowing at the individual level and at the zip code level, showing that though mortgage balances grows more in areas with a larger fraction of subprime borrowers, within those areas, debt growth is driven by high credit score borrowers.

Using zip code level data, Mian and Sufi (2009) show that during the period between 2001 and 2006, the zip codes that exhibited the largest growth in debt were those who experiences the smallest growth in income. They argue that the negative relation between debt growth and income growth at the zip code level over that period is consistent with a growth in the supply of credit to high risk borrowers. We show that this negative relation does not hold for individual data. The differences in credit growth between 2001 and 2009 are positively related to life cycle growth in income and credit scores. Moreover, debt growth for young/low credit score borrowers at the start of the boom occurs primarily for individuals who have high income by 2009, and the growth in income is associated in a growth in credit score.

The positive relation between income growth and debt growth during the credit boom casts doubt on the notion that there was an increase in the supply of credit, especially to high risk borrowers. Instead, it is more likely that the rise in house prices caused an increase in mortgage balances. This is confirmed by the fact that the fraction of borrowers with mortgages did not rise for any quartile of the credit score distribution(.)

(T)hough the fraction of investors with prime credit score is very similar across quartiles, in quartiles with high share of subprime, investors exhibit larger increases in mortgage balances during the boom and a more severe increase in foreclosures during the crisis. This difference in behavior for prime investors may be driven by the behavior of real estate values.
(KE: In other words, investors who were prime borrowers were a large source of delinquencies in zip codes with a high proportion of subprime credit scores.  So, more volatile prices probably were the cause of higher default rates in those zip codes.  My work has shown that those zip codes were located in specific areas, and the volatility comes from a systematic behavior of Price/Rent that is probably unrelated to credit markets.)

We find that most of the increase in mortgage debt during the boom and of mortgage delinquencies during the crisis is driven by mid to high credit score borrowers, and it is these borrowers who disproportionately default on their mortgages during the crisis. The growth in defaults is mostly accounted for by real estate investors.  


  1. Well, I will have to read this post a couple more times and the Albanesi paper more throughly. Looks right.

    I will note this: The words "property zoning" are entirely absent from the Albanesi paper. I did a word search.

    In fact, even those papers I agree with, such as Ferrero's work on trade deficits and house prices, never mention "property zoning."

    I would think at this late date at least a sentence or two would be warranted….


    fun chart

    1. Yeah. They do some fun stuff over there. They have similar charts of the S&P 500 value vs. dividends that are interesting. They allow you to really see periods of upheaval vs. normalcy.

  3. Well, if anyone cares, I read the Albanesi paper. I think it holds water.

    But, it is a feature of modern-day macroeconomics that 1) good or bad or mediocre papers will be ignored if they undercut conventional wisdom 2) A reader might agree with a paper, but barring hours and hours of legwork, have no real way to check the paper.

    For example, "Our analysis is based on the Federal Reserve Bank of New York Consumer Credit Panel/Equifax data, a large administrative panel of anonymous credit files from the Equifax credit reporting bureau."

    Obviously, I am never going to check their use of the Equifax data. No one else is either. There was one example of some University of Massachusetts guys really checking some economists work, and finding serious shortcomings. But you have to get down in there with the data, and spend days. Whenever I have checked social science work, I usually found flawed premises, convenient start and finish dates, and dense calculus that could be rendered meaningless by some inaccuracy in the data. Sometime there is obvious shaving of data going on.


    That said, If the Albanesi paper holds water, it indicates the way the US has combatted high house prices was to cut middle-income and below people off from credit markets, though they were not the cause of the "problem."

    The issues of property zoning were not even mentioned, let alone addressed, nor the issue of current account trade deficits and house prices.

    It is as if the propertied-financial class declared class war on Ordinary Joes trying to buy a house.

    As an aside, before 2007-9, buying a house was statistically a very safe investment, as house prices (in the aggregate) rarely fell. That is one reason that AAA securities sold well, is that it seemed impossible that a general housing correction would be anything but minor.

    The Fed and national policies made sure 2008 would be different.

    Getting back to this:

    The above chart shows that in 2000, a median priced new house was about 4 times annual median household income, but by latest reading that ratio had hiked to 5.4 times.

    I do not know why they used new rather all median-priced home sales. But anyway, it appears housing is becoming more expensive, relative to household incomes.

    Is that worrisome?

    1. Really, there are many reasonable policy positions that could lead to various types of housing markets. Germany and Switzerland, for instance, don't encourage owner-occupied housing so much, so in those countries, households tend to consume less housing and prices aren't as high. That's fine. If we want to have an economy where there is less housing consumption, that's a reasonable position to have. The problem we have is that we are imposing policy regimes that we don't even understand. We are still encouraging homeownership, but we are imposing non-price obstacles so there is a severe class-based effect where households with income and wealth gain advantages and households with lower incomes and less wealth are left with rising rents. And, everyone is beating the bushes to find the causes of how the rich are gaming the system, when the system is being gamed by the GSEs and CFPB.

  4. Since no one seism to have noticed---PCE core July at 1.4% up YOY, down from 1.5% YOY up in June.

    Will the Fed tighten even as PCE core drifts south from target? And is 2.0 % a target or a ceiling?

    1. It seems like we are likely to get pulled into a contraction by tight policy. I would guess that either inflation continues to decline, and the Fed holds rates level for too long, or we get a bit of a spike in inflation from temporary noise, which the Fed takes as justification to raise rates again. I don't see much chance for them to avoid being too tight.