Quantcast
Partner Insights

Digital Risk Explains Understanding Credit and Manufacturing Risks

MAR 21, 2014 1:43pm ET

Once upon a time an individual’s credit rating was the biggest concern for mortgage lenders. If an individual had good credit and steady income, everyone was happy. The individual got the home, the lender got the sale, the mortgage was a done deal, and everyone lived happily ever after.

Unfortunately, or perhaps even fortunately depending on your perspective, we can still live happily ever after, we simply must work a bit harder to navigate through new industry dynamics and embrace the changes that are a large part of our new reality. Industry happenings surrounding the housing crash and new QM origination rules force us to look differently at our loan manufacturing process.

What do we mean by manufacturing process? The primary components of the loan manufacturing process are obtaining the borrower’s application, collecting a variety of information from the borrower, obtaining appraisals and other information about the property, submitting the information to underwriting for review, approval decision, and final funding.  Completing a successful manufacturing process includes analyzing the borrower’s level of credit risk and the lender’s willingness and ability to assume that level of risk. Lenders rely on various tools and analytical methods to determine the borrower’s default potential, or credit risk.

Prior to the housing meltdown and lurking in the background; however, has been another type of risk –manufacturing risk. The determination of the loan as being corrupted or misrepresented is referred to as the manufacturing risk. In other words, the information value of such data as borrower’s income, debt to income ratio, loan to value ratio and/or appraised value, have been weakened, eroded, corrupted and/or in some way misrepresented.   

When the information value of these key data elements are reduced in some significant way, any analysis or projection of the future performance of an infected loan is suspect; as it is accomplished under false premise.  When this occurs, the likelihood of risk is underestimated, the loan is overpriced and it underperforms relative to investor expectations.

Today’s manufacturing process weakens the value of key information elements in numerous ways.  One key information element is the appraisal.  Our research has shown that appraisal values can contain substantial volatility, even if accomplished by a licensed appraiser in accordance with USPAP guidelines.  Our research has shown that the 95% confidence limit on appraisal values is +/- 15% of appraised value. 

That means that qualified appraisers could differ substantially and still be rendering appraisals that would meet industry standards.  For example, a home with a “true” market value of $300,000 could be realistically appraised at a value anywhere from a high of $345,000 (+15%) to a low of $255,000 (-15%) and still be considered a “valid” appraisal, that is one that was within the 95% confidence limits.  When an appraisal is within the confidence limits, the value generated can be justified given an analysis of relevant comparables in accordance with standard industry practices. 

The range of acceptable values (those within the confidence limits) are as broad as they are due to the degree of accuracy and consistency present in today’s appraisal process, which has a great degree of subjective (appraiser) input and leeway. Further, given this loose standard, an analysis of loans submitted for funding and/or purchase shows that more than 10% of the valuations submitted evaluate the subject property at values in excess of 15% above true market value, further reducing the information value of the appraisal.

As can be seen, the idea that the appraised value is foolproof and can be relied upon without question is an idea that is easily challenged.  Moreover, that means that published Loan to Value (LTV) ratios are subject to substantial volatility, often with the originator, the issuer and/or the investor being aware.

A second area where the information value is suspect is the estimation of income.  While there has been a move away from “Stated Income, Stated Assets” (SISA) loans, this has not necessarily eliminated the volatility of income, nor has it eliminated the accuracy of a Debt To Income (DTI) statistic, one that is relied upon heavily by QM standards, Dodd Frank and Ability to Repay (ATR) guidelines.  There are a number of issues.  When the IRS is used, via 4506T processes, the number that is received is historical, relating to the most recent tax filings, which may or may not reflect the borrower’s current income. 

Moreover, many borrowers have more than one kind of income, such as investment and rental income, as well as commissions and bonuses, along with the standard W-2 pay stub information.  Given these many sources of income, there are a number of issues.  One is accurately aggregating all sources of income from a given borrower; often a difficult task.  The other is the calculation of income from multiple sources; often the weighting of income varies according to source.  For example, while W-2 (recent pay stub) information is weighted at 100%, income from rental properties is weighted less, often at only 50%.

As can be imagined, the process of finding and aggregating all of the sources of income across a borrower’s portfolio often results in inaccurate information and reduced information value of the income statistic.  Further, the application of the appropriate weighting formulas also introduces additional volatility resulting in reduced information value of income.  As a result, underwriters seeking to meet QM guidelines, which specify a maximum DTI of 43%, often employ a “practical” maximum of 42% or 41%, in an attempt to account for income volatility and avoid breaking ATR guidelines of 43% DTI.   Should these guidelines be breached, the borrower has recourse upon foreclosure, asserting that the lender did not fully bet his/her ability to repay, thus undermining any potential foreclosure proceedings.

Chief Analytics Officer, Digital Risk LLC