16454.jpg

The use of analytical tools to mine data has increased in popularity and effectiveness in recent years. Many industries have transitioned from highlighting data quantity to emphasizing data quality. Just ask Billy Beane, general manager of the Oakland Athletics, who used analysis - part of which included evaluating unconventional data, such as on-base and slugging percentages - to select the strongest lineup of players that would provide his team with the best chance of winning. Using this unique mathematical approach to drafting and identifying key draft picks and free agents, the Oakland Athletics finished the 2001 season first in their division, winning 20 consecutive games during one impressive stretch.

In baseball, the number of home runs that a player hits is not the sole indicator of his greatness. Equally, servicers cannot rely on a borrower’s credit score by itself as an effective predictor of long-term performance. Understanding which borrowers may present more risk is now becoming an area of increased focus. The quantity of data is outweighed by the quality of the insights that exist within, but sifting through petabytes of data isn’t impressive unless you’re finding quality information and making it actionable.

 

Predicting default risk through analytics

It wasn’t until a few years ago that credit history and performance were the primary indicators of default risk. We have come a long way in a very short period of time, and now, as in baseball, we can leverage unique insights to more accurately gauge a borrower’s propensity to default. These insights can be found from the following three primary sources.

1. The loan: As Patrick Barnard, editor of Servicing Management, stated in March, “Default servicing is here to stay” - citing poorly underwritten “legacy” mortgages as a driving force of elevated foreclosure rates. To his point, one of the elements that predict default risk is in vintages. Less stringent underwriting standards coupled with falling home prices contributed to higher delinquency rates seen in mortgages originated in mid-2000. In fact, the 60-day delinquency rate for mortgages originated in 2006 and 2007 is twice as high as vintages from 2005 and earlier. The elevated delinquency rate includes borrowers with credit scores above and below 660 - evidence that it takes more than evaluating a borrower’s credit score to gauge default risk.

2. The borrower: There is a clear correlation between how much someone earns and how likely he is to become delinquent. According to data collected through The Work Number, consumers earning less than $30,000 annually are four times more likely to become 60 days past due than those with incomes of $100,000 or more per year.

Gainful employment is important, but it is not just about having a job. Consider how a worker is paid. Consumers with salaried positions are more likely to stay current on their mortgage payments. The benefits of a salary allow for better planning because the amount received is consistent and supports the ability to better allocate financial resources. Conversely, hourly workers are more risky across all income bands. This is primarily due to the lower amount received over all, as well as the fluctuations in income due to inconsistent hours.

smcover1506.jpg

Similarly, job tenure is a very good predictor of payment ability. Equifax performed a study that looked at delinquency rates tied to job tenure and borrowers’ home equity line of credit (HELOC) balance. The results showed that consumers who stick with one employer for an extended period of time are more likely to stay current on their HELOC payments than those with less than one year of tenure. Overall, consumers who stick with one employer for an extended period of time are more likely to maintain regular mortgage payments. Conversely, consumers who have become inactive very recently (within six months) are at a higher risk for default and delinquency. Analytical tools, such as employment alerts, can be used to perform outreach to customers at very early delinquency stages to gain insight into any potential life changes. Reaching the homeowner earlier in the process increases the chance of borrower response or engagement.

3. The property: For servicers, determining occupancy is a primary way to mitigate risk during the acquisition of mortgage servicing rights. In addition, it ensures due diligence and minimizes putback risk when acquiring new loans. A borrower’s intent to occupy a subject property can be materially misrepresented in the loan process. In 2011 and 2012, approximately 14% of loans sold to Fannie Mae included occupancy misrepresentation. This misrepresentation occurs because mortgages for owner-occupied properties have lower rates and less stringent requirements. Due to the high risk associated with non-owner-occupied properties and mortgages, some mortgage insurers will only underwrite policies for owner-occupied properties.

Non-owner-occupied properties have a propensity to default for a number of reasons. If the property is being rented, a steady stream of income to pay the mortgage may not always be available. The borrower may own a second property with a mortgage payment due.

The condition of the property may also be affected based on whether the occupant is the owner or simply a renter. Without much of a financial stake in the property, a renter will have less incentive to provide proper care and upkeep. Assessing the occupancy status of a property is possible, but most servicers do not have the tools to do so. Relying on the mortgagor to provide documentation increases this risk - which does not guarantee the accuracy of the information and increases the risk of default and repurchase.

Further, investors (including the government-sponsored enterprises) and mortgage insurance companies are faced with loans that may become or have already become delinquent. These loans need to be returned to the originator, as they are not compliant with reps and warrants. The repurchase and rescission process is expensive and affects a firm’s bottom line.

Today, identifying owner occupancy consists of leveraging credit header data and public record data, which is often insufficient or has a significant time lag in update cycles. Instead, the focus should be on verifying whether a property (single-family, multifamily, condo, co-op or apartment) is occupied by the owner of record at originations for limited uses cases, such as refinances and loan modifications, after the loan has been assumed by the investor or covered by the mortgage insurer. In order to determine ownership status, servicers should cross-reference multiple addresses and independent data sources, including the following:

 

Turning data into dollars

The in-depth empirical research of statistics in Major League Baseball is called Sabermetrics, and it has revolutionized the way in which players’ skill and impact are measured. And though it lacks a quirky title, the approach to data analysis in the mortgage industry is similar and equally powerful.

Accurately measuring default risk goes a long way toward mitigating portfolio risk. Today, originations are healthier due to the stringent guidelines imposed - a primary result of the recent housing crisis. It’s turning out to be a great year for mortgages; however, there is still plenty of work to be done internally, and servicers must start by ensuring they have access to dynamic analytical tools and that they understand how to appropriately leverage them. If done properly, poorly underwritten “legacy” mortgages can be tossed, which would make room for more profitable loans - turning a pop fly into a grand slam.

 

Rosie Biundo is senior director of product marketing for Equifax Verification Services. She can be reached at rosa.biundo@equifax.com.

Default Risk

What Do The Mortgage Industry And Baseball Have In Common?

By Rosie Biundo

Hint: In both, risk is becoming an area of increased focus.

 

coverphoto.indd

 

 

 

sm_bod

sm_bod_i

sm_bod_b_i

sm_bod_b

hyperlink

sm_cal_date

SM_text

SM_caption

sm_subhead

SM_F_subhead

 

 

 

SM_F_1stpara

 

sm_bod_last_graph

SM_last_graph

 

SM_pullquote_text

SM_pullquote_text

sm_s_h

sm_s_h1

sm_names

 

 

SM_frontoffice_1stpara

 

 

 

SM_Frontoffice_Dropcap

Sidebar Headline