Mortgage Market Risk Models Built on Flawed Assumptions

by devteam October 2nd, 2010 | Share

“Over the past twenty years, understanding of and business practice in mortgage markets has been influenced significantly by the application of statistical models. Mortgage underwriting was automated using statistical models of default and default loss, and statistical models of denial rates and loan pricing were used to test for discrimination in lending. Efforts to measure mortgage market discrimination and credit risk have been propelled by an increase in the loan-level data available through various resources. Unfortunately, as researchers strived to produce results from these data, critical statistical errors were overlooked and then repeated in what has become the “conventional approach” to measuring discrimination and credit risk.”</p

That is the executive summary of A Review of Statistical Problemsrnin the Measurement of Mortgage Market and Credit Risk, a paper sponsored by MBA’s ResearchrnInstitute for Housing America (RIHA) and conducted by Professor Anthony M.rnYezer of George Washington University.   The study examined three models used inrnconventional approaches to testing for discrimination in the granting ofrnmortgages – testing based on applicant rejection equations, based on mortgagernpricing equations, and based on mortgage default equations and one conventionalrnapproach to measuring credit risks. </p

The study found that statisticalrnerrors had been perpetuated through many different analysis of the underlyingrndata, those errors were then unknowingly institutionalized and accepted as conventional practice, i.e. some simplernstatistical models are not firmlyrngrounded in economic theory.</p

According to Yezer, the major simplifyingrnassumption made in the models is that borrowers have no knowledge of thernmortgage lending process and do not select mortgage terms strategically.  The theory has been that the lender selectsrnthe mortgage terms and borrowers are oblivious to the effects of their own decisionsrnon the transaction.  For example,rnconventional statistical techniques assume that borrowers determine the amountrnthey are willing to pay for a home without considering that this decision willrnimpact on the chances they will be rejected. rnThis, the author says, flies in the face of reality.  Any buyer finds out early in the process thatrnhis ability to qualify will be based largely on the price of the home he selectsrnand the amount of his down payment.  Thisrnself-selection is largely overlooked in conventional studies.</p

Another problem with conventionalrntechniques is omitted variable bias.  Forrnexample, in an evaluation of discrimination based on rejection, the variablernindicating race made be analyzed incorrectly as correlating with rejection whenrninstead it is a variable that is a dummy for race (perhaps a lower averagernincome or credit score for a minority population) that is causing the effectrnbut is not included in the equation.   </p

Incorrect coding also led tornincorrect interpretations.  For example,rnin an analysis of Boston Federal Reserve data, one study found a significantrnsource of errors was the difference between what was initially claimed by thernapplicant and the final determination by the underwriter.  Sometimes both the initial false claim andrnthe verified information were both retained and coded and it is unclear whichrnshould be properly included in a statistical analysis.</p

The complexity of the loanrndecision leads to other problems.  Forrnexample, some characteristics of the borrower may become less important whenrnthere is a cosigner.  If there is a poorrncredit history, then LTV takes on more significance.  When one study interacted variousrnunderwriting variables with a race dummy, it was found that, for somernvariables, being a minority was an advantage while in others it was arndisadvantage.</p

These types of complexity alsornaffect analysis of loan pricing as a determinant of discrimination.  Loan terms used to determine risk are notrnonly causes of APR, they are caused by APR. rnFor example, while the likelihood of prepayment may cause higher APR, itrnis equally true that higher APR makes prepayment more likely.</p

The report states that thernrecent financial crisis revealed many shortcomings in the market, one of whichrnwas that default and default loss models “woefully underestimated creditrnlosses.”  The two common models usedrnare ex-ante default based on seasoned mortgages and relying on the probabilityrnof default falling drastically with time. This model does not take intornconsideration borrower demographics and eliminates the chances for statisticalrndiscrimination as lending decisions are made on objective criteria.  The second model is designed to estimate therncash flow from mortgages and includes variables reflecting conditions at applicationrnand those reflecting the evolving conditions of the mortgage and housingrnmarkets.  This model is also flawedrnbecause the population surviving into out years is fundamentally different thatrnthe original population of mortgages simply because they have failed to prepayrnor default.</p

The study concludes thatrnthe outcome of a mortgage transaction involves the simultaneous considerationrnof many factors and this complexity is ignored by current theoretical modelsrnthat consider two variables at a time. The serious limitations of currentrnstatistical approaches to testing for discrimination and credit risk inrnmortgage lending have likely contributed to recent problems in mortgage markets.rnIf these limitations are not recognized and naïve reliance on them continues,rncurrent problems are likely to recur in the future. Alternatively, there arernmajor gains to be made if economic analysis of mortgage market discriminationrnand mortgage credit risk can be improved.</p

Common Sense Not Found in Automated Underwriting Engines </p

Expanding the Pool of Eligible Homeowners: Common Sense Underwriting Needed

All Content Copyright © 2003 – 2009 Brown House Media, Inc. All Rights Reserved.nReproduction in any form without permission of is prohibited.

About the Author


Steven A Feinberg (@CPAsteve) of Appletree Business Services LLC, is a PASBA member accountant located in Londonderry, New Hampshire.

See all blogs


Leave a Comment

Leave a Reply

Latest Articles

Real Estate Investors Skip Paying Loans While Raising Billions

By John Gittelsohn August 24, 2020, 4:00 AM PDT Some of the largest real estate investors are walking away from Read More...

Late-Stage Delinquencies are Surging

Aug 21 2020, 11:59AM Like the report from Black Knight earlier today, the second quarter National Delinquency Survey from the Read More...

Published by the Federal Reserve Bank of San Francisco

It was recently published by the Federal Reserve Bank of San Francisco, which is about as official as you can Read More...