One model got it right. New consumer behavior is likely an adaptation to the FICO risk scores.

February 5, 2008

Comment on ‘Last year’s model: stricken US homeowners confound predictionsBy Krishna Guha and Gillian Tett The Financial Times, Comments and Analysis, January 31, 2008 19:01

Story The Financial Times raised an interesting question this week about changes in consumer repayment behavior, and the failure of mathematical models to keep up to these changes in relation to the subprime mortgage mess. It would seem that a statistically visible number of consumer have been opting to continue repaying their credit card bills and car loans even while defaulting on their mortgages. This contradicts conventional wisdom which suggests that rather than risk foreclosure, households would prioritize the home over small debts, paying mortgage payments first and foremost when in financial trouble.

Malcolm Knight, head of the Bank of International Settlements, summed up this new pattern of repayment as follows (quoted by Guha and Tett):

“Now what seems to be happening is that people who have outstanding mortgages that are greater than the value of the house, or have negative amortization mortgages, keep paying off their credit card balances but hand in the keys to their house… these reactions to financial stress are not taken into account in the credit scoring models that are used to value residential mortgage-backed securities.”

The article goes on to suggest two possible reasons for the change in behavior that escaped the mortgage default models: first that it may be due to cultural changes that lessen the stigma associated with missing a payment or loosing a home, and second that people may no longer have an incentive to pay mortgages where the loan-to-value ratio has become excessively high with dropping property prices. That is, they’ve decided it’s just not worth it.

Implications Drawn to its logical conclusion, what this piece implies is that on a large scale the American consumer no longer minds having their property taken away from them and might even willingly abandon it once they’ve calculated that it’s too expensive. Hmmm. That sounds kind of… doubtful. Consider the gushing tears and suicidal thoughts of precarious homeowners featured so prominently in Scurlock’s (albeit melodramatic) documentary, Maxed Out. It’s unfortunate that these financial journalists, who appear to live across the pond in London, only had access to macro data. If they had had the chance to come over and investigate the actual practices of American consumers up close, they might have considered dropping culture and economic rationality – two of the falsest friends the social sciences have ever confabulated – and discovered some more plausible reasons to account for this new consumer behavior. Hint – it’s a risk model. FICO® consumer credit bureau scores which receive so much attention in consumer circles – and almost none in the financial press related to the mortgage crisis – are one of the key pieces of information used for matching loan products to consumers in the U.S. In 2001, after pressure from consumer groups started to build in Washington, the scores were released to the public which is now able to purchase access to their scores (see www.myfico.com). So consumers know a thing or two about how the models work, and there is plenty of advice in circulation to tell them how to behave accordingly. What is ironic is that the scores only came to public attention after they were adopted by the Government Sponsored Agencies (GSEs) in 1995 as a sub-component of their automated underwriting programs (Loan Prospector® at Freddie Mac and Desktop Underwriter® at Fannie Mae). From there they worked their way through the mortgage industry into the securities underwriting models (such as S&P’s LEVELS®). Interestingly enough, no mortgage data used to calculate FICO® scores, which were originally designed as risk indicators for small consumer credit, supporting in particular the credit card industry. They were never redesigned to accommodate the mortgage markets because the bureaus have traditionally not had access to mortgage data.

Conclusion Since these scores are the obligatory passage point to further consumer credit and play a role in refinancing – i.e. getting out of a subprime loan, getting another mortgage… and so on – a move to preference credit cards payments over home loans would probably not have much to do with a growing indifference towards foreclosure. Rather it would be a performed consumer response targeted at protecting their precious risk scores. Not convinced? Remember the Paulson Plan released in December? It suggested interest rate freezes on ARMs but would have limited these to cases where the borrower had a FICO® of 660 or more. This means that the category of people the federal government will agree to rescue pay their credit card bills faithfully and on time… even when they can’t afford their mortgages. In this light, paying credit card bills would be a way of waving a white flag that cries out ‘help me (I’m helping my self)’, and not at all a way of bailing out of an overpriced home. If we can consider changing consumer behavior a form of getting it right, then at least one risk model didn’t get it wrong… At least, not this time.

11 Responses to “One model got it right. New consumer behavior is likely an adaptation to the FICO risk scores.”

  1. yuvalmillo Says:

    Thanks! This story of ‘playing the models’ is a very interesting one. Such phenomenon can help to generalise from this case to other fields, financial and non-financial alike. One thing you may want to address is the degree to which the FICO model is indeed a critical point of passage. Over here, in the UK, a similar metric calculates the loan-risk associated with individuals. Yet, creditors declare that they do not use this model as an exclusive criterion and take into account data from many other sources. I have no concrete evidence to what degree this does happen, but it would be interesting to look into that.
    Another issue that’s implied here is the one-model risk. If creditors tend to rely on a single model and the there is some event that makes a large chuck of prospective loans seem risky, then the credit market may dry up. Of course, a mainstream economist may say that at this stage other actors would step in and supply liquidity, but as we know from other instances of ‘flight to liquidity’, this may not happen.

  2. Alison Kemper Says:

    Although I don’t live in the US, I read their blogs. Many have suggested that the pattern of repayment is due to the bankruptcy act “reforms” passed over the last couple of years. People are tied for life to their school loans and credit card debts.

    However, legislators did not anticipate that it would be worthwhile to walk away from your house and mortgage, so this category of debt is now much more escapable than consumer debt. When the value of the house dips below the amount of the high-ratio mortgage, Presto! It’s time to walk.

  3. marthapoon Says:

    It’s quite hard to convey to people who have not dealt with the US consumer credit system what the FICO is. It is not the only consumer risk metric available on the market – each of the three major credit bureaus sells many score products — but it is considered the ‘gold standard’. It is the only bureau score used in mortgage underwriting since the industry followed the lead of the GSEs in 1995; and most importantly it is the one that the public has access to.

    The success of the FICO is a business fiat has had tremendous consequences on the public perception of financial risk. The appearance of an industry standard has to a large extent de-problematized the idea of attaching ‘a (singular) risk’ to individual consumers. While multiple rankings would raise questions about the nature of calculation the public as mostly focused on accuracy – is the data used in ‘the’ calculation correct. (Espeland and Sauder have a nice article discussing the differences between single and multiple rankings law school and business school rankings as contrasting cases.)

    [http://papers.ssrn.com/sol3/papers.cfm?abstract_id=862924]

    Without actually going out and doing empirical research it would be impossible to know whether people are ‘willingly’ (which implies to some degree, coolly) giving up their homes, or as I suggest, struggling to continue paying their credit card bills at all costs even in extreme financial distress to protect their scores. But what seems certain, and in response to the FT’s claim that ‘risk models got it wrong’, is that the FICO has participated in the emergence of this new behaviour – the shift towards privileging credit card over mortgage repayments.


  4. […] One model got it right. New consumer behavior is likely an … Conclusion Since these scores are the obligatory passage point to further consumer credit and play a role in refinancing – ie getting out of a subprime loan, getting another mortgage… and so on – a move to preference credit cards … […]

  5. danielbeunza Says:

    A fascinating explanation for the meltdown. And a great illustration of the value looking at models — and of the social studies of finance.

    I wonder what the implications are… does it imply that several measures of risk are better than a single one? or that the one single measure of risk should be more comprehensive?

  6. Zsuzsanna Says:

    Martha,

    Thanks for providing a well-founded alternative hypothesis to counter FT’s surprisingly simplistic assumptions.

    Your explanation is based on acknowledging that consumers are attached to their homes (which the FT explanations simply give up), but while asserting that this is not a purely “emotional” relation: they are also calculating. In fact consumers are knowledgeable about how they are being calculated by others (by the government and the credit bureaus), and take that into account in their calculations.

    Of course my first reaction was what you also said later: it would be nice to confirm it empirically by getting data on consumers’ calculations.

    One clarification: is your conclusion that consumers are in effect, not risking foreclosure (by the seemingly counterintuitive pattern of defaulting on mortgage rather than the kind of debt they know counts in credit scoring, which is the basis of the bailout)?

  7. Chase Says:

    I used information from that blog its great.Excellent blog, added to favorites!!

  8. typewritten Says:

    Chase,

    You’re a credit robot. Get lost.


  9. […] make decisions, and of the material devices that they use to accomplish them, in the fashion of Martha Poon and Joe Deville. Possibly related posts: (automatically generated)Wising UpNo TitleThe State of […]


  10. […] rid of automated underwriting and the use of credit scores to underwrite FHA loans. The lax approvals regularly defy common sense and have been a PRIME […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: