What went wrong with subprime loans in 2006?

Yuliya Demyanyk and Otto Van Hemert report:

We analyze the subprime mortgage crisis:
an unusually large fraction of subprime mortgages originated in 2006
being delinquent or in foreclosure only months later.  We utilize a
loan-level database, covering about half of all US subprime mortgages,
and identify two major causes.  First, over the past five years, high
loan-to-value borrowers increasingly became high-risk borrowers, in
terms of elevated delinquency and foreclosure rates.  Lenders were aware
of this and adjusted mortgage rates accordingly over time.  Second, the
below-average house price appreciation in 2006-2007 further contributed
to the crisis.

Neither point is shocking news, but the mystery deepens upon inspection.  After 2006, 2001 was the next worst performing year for mortgage repayment, a puzzling fact.  I had expected a rising crescendo of failures; why did things suddenly get so much worse when they did?  Furthermore variable rate loans and low documentation loans do not seem to worsen disproportionately in 2006, contrary to common suppositions.  The ratio of loan value to house value is a critical variable, but again there is the puzzle of why 2006 was so much worse for these loans than preceding years.  Unless you think everyone is "flipping," (not the case), falling home prices don’t stop you from repaying your loan.  Note also that lenders could see the risk of these loans worsening, and it was reflected in the rates they charged, although apparently not enough.

Addendum: Here is James Hamilton on same.


This paper will be quite influential if only beacuse of their very large dataset.

I view the entire subprime crisis as originating from one single factor- people thought significant HPA could go on forever. From the blog Calculated Risk, it's obvious how many prime borrowers used subprime loans(typically the ARM resets) as a bridge loan for real estate speculation. Securitization imposed a moral hazard problem in the chain. Note that the authors say that rates were increased ex-ante to reflect the risk but now ex-post it's obvious it wasn't enough.


The article appears to be talking about loans that originated in 2006.

I think it's clear that many of those loans were fraudulent, that by 2006 crooks and con-artists had discovered how disturbingly easy it was to get money they never intended to repay. Mortgage brokers still get their commissions, even if the "borrower" is fictitious.

That's the trick of the data, it was originations that occured in 2006 that have been the problem, not originations from 2004 that reset in 2006. ARMS originated in 2004 remain at very low delenquencies through 2006. 2006 originated loans are worse than all other years (beginning in month 1-2 in most cases), but no category of them (ARM, lo doc, etc) is singularly responsible for the increase.

It looks to me like, 2006 loans were made with apprased value much higher than actual value. Which is generally true as every boom turns to a bust.

2006-7 was the breaking point from a virtous cycle to a vicious one.

Very low interest rates in 2002-3 and their slow rise inflated house prices - since the primary component of house payments is interest, house prices act like bonds and rise when rates fall, all else equal. Rising prices allowed zero downpayment purchases and rapid refinancing with equity extraction, improving borrowers' cash flow and credit profile. Mortgage products (i.e. teaser rate arms) that kicked the cost of credit forward past a likely refinancing expanded, and subprime expanded tenfold, representing 40% of mortgage originations.

Stable home prices and rates (notably a flat curve) in 2006 stopped that cycle. Now, flat-to-down home prices stop refinancingn and equity extraction and teaser rates expire, both hammering borrower cashflow. Overbuilding and stuck markets leave a higher percentage of homes empty than ever recorded. Mortgage credit distribution was built on rating agency assumptions that were just insufficient, leading to massive withdrawal of funds from the market. The weakening dollar keeps rates high. All this increases homeownership costs, further depressing prices and continuing the vicious cycle.

I'd like to see analysis of broader data sets about how consumers allocate their money to housing relative to other expenses. There's a lot of focus on all aspects of mortgage lending, but I'd like to see more on overall household decision-making.

The recent news story about consumers paying credit card bills before mortgages showed a shift in behavior that might be explained by the bundling of transaction services (the card as a cash substitute, now adopted by many as every day behavior) with credit in the credit card product. With increasing consumer debt, do people see mortgage debt differently in the past 5 to 10 years?

I also strongly believe that for the current system to function well there must be intensive prosecution of fraud and misrepresentation among lenders, as well as stringent ethics policies on lenders. As a mortgage professional as well as a borrower myself, I've been personally horrified by some of the practices I've seen and experienced. The system is only as good as its ethical foundation.

falling home prices don't stop you from repaying your loan
People may see it as throwing good money after bad. Whereas when you are in the black, you want to protect the expected profit.

Keep in mind I am not trying to explain the whole effect, just a possible small addition.

2006 vintage loans have an unusually high default rate for primarily one reason: fraud. When the pace of home sales started to slow down in late 2005 the lenders responded by dropping their lending standards so low that commissioned brokers essentially invited fraudulent loan applications. There was never any intention by the "borrowers" to pay those mortgages. Look at the story of Crisp & Cole in Bakersfield CA if you want a case study.

I think Dude may be on to something beyond the speculation of fraud. From my experience in the industry, by 2006 a lot of people were simply expecting to get the highest LTV that they could and thus a lot of loans were made with little equity.
Add to that the timing. Late in the market is when a lot of people who shouldn't buy do because 'everyone' is. Not all of them achieve this through fraud, but brokers are a lot more willing to shepherd someone through a credit fixing process when they are running out of other prospects. However, this only fixes the credit scores, not necessarily the bad habits that created the original bad credit.
Finally, compound that with desperate mortgages brokers and lender account managers (also commission based) and the line between fraud and negligence gets very thin.

I spoke with one broker in the Ann Arbor, MI market and he told me that 20% of his recent closings featured sellers who had to bring money to the closing to pay off their loan.

I can imagine that by 2006 there were a lot of buyers who never really had great credit to begin with, had always been renters, and who when faced with having gotten themselves in over their heads just said "screw it" and walked.

Can anyone point me to a place that lists which state's are mandatory non-recourse and which aren't? I'm most interested in Michigan, but I'm curious about other states as well.

Comments for this post are closed