Inventory
info icon
Single family homes on the market. Updated weekly.Powered by Altos Research
722,032+456
30-yr Fixed Rate30-yr Fixed
info icon
30-Yr. Fixed Conforming. Updated hourly during market hours.
6.97%0.00
EconomicsLegalServicing

Living large

More than 1 million loan modifications have been attempted since the housing meltdown, which offered borrowers billions in total payment reductions.  Yet, modification redefault rates have been excessive; some populations exceeded a 60% redefault rate within the first year.

This begs the question: Where did the money go? How did borrowers use those billions in payment reductions? If a borrower had a $3,000 monthly mortgage payment that was reduced by 20%, that borrower had an additional $600 in discretionary cash per month, or $7,200 per year, with which ensure the success of the mortgage.

For every 100,000 modifications, an estimated $240 million in discretionary cash was offered to the borrower community for each year the modification was in effect:  ($200,000 loan, 6% interest rate, 20% payment reduction). Given that more than 1,000K loan modifications have been attempted, the borrower community has had on average $2.4 billion in discretionary cash per year to shore up mortgages. Yet, apparently many did not.  

Some argue that this was a case of insufficient payment reduction. Others argue that there may also be a pronounced effect of negative equity; that is if the bank had offered principal reduction, the redefault rates would have been lower.  

FACTORING IN LIFESTYLE

While the preceding items are relevant, we have discovered a factor that is more pivotal than loan modification terms. That factor is lifestyle.  
The lifestyle of the borrower receiving the loan modification plays into the how the borrower is likely to use the discretionary cash. Will that extra cash go toward the mortgage? Or, will that money be used to cure other problem debts, of which there were many in this population? It is important to realize that a borrower receiving a loan modification almost always has other debt problems, making that borrower inherently more risky.

The loan-modified borrower has a higher problem-debt percentage relative to the typical borrower and more debt problems than just the mortgage at the time of loan modification.  
Exhibit 1: The Profile of a Borrower at the Time of Loan Modification, reveals that in 2009, borrowers who were granted loan modifications had numerous seriously-delinquent credit relationships, as measured by the metric:  problem-debt percentage.  The problem-debt percentage is the percentage of the borrower’s credit lines that have been 60-plus days past due within the last 24 months. (Please see flip book under Magazine All Access tab to view charts.)

This finding was based upon a random sample of 1,000K borrowers from the 2009 time frame that was drawn from a population that was weighted toward subprime, Alt-A and prime jumbo.  However, almost 50% of this population also contained prime conforming loans.  

The Exhibit 1 data reveals that the in 2009, the average borrower (group:  ALL_2009) had a problem debt percentage of 16% — that is 16% of this borrower’s active credit lines had been seriously delinquent within the last two years (2007 – 2009, inclusive).

However, when you compare the average borrower’s performance (ALL_2009) with the performance of the modified borrowers, it is clear that the modified borrower in 2009 had a significantly more problem debt than the average borrower.  For those borrowers who qualified and accepted a loan modification in 2009 (LOAN_MOD_2009), they had a problem-debt percentage of 30%, significantly higher than the typical borrower (ALL_2009, 16%).  Moreover, those borrowers who failed at loan modification by 2011 had significantly higher problem-debt percentage ratings in 2009:  LOAN_MOD_2011_Fail had a problem debt percentage of 51% in 2009.  Those who succeeded at loan modification through 2011 (LOAN_MOD_2011_Succeed) had a significantly lower problem-debt percentage in 2009.

Clearly, the status of credit relationships beyond the mortgage had significant influence on future loan mod success.   Those borrowers who succeeded at loan modification went into loan modification with a smaller portfolio of debt servicing problems than did those who failed.  And, loan modification policy officials would be well advised to consider a broad array of debt servicing issues, such as problem-debt percentage as they adjust their loan modification policies to ensure higher success rates.  

As is shown by Exhibit 1, the discretionary cash created by the loan modification has many potential applications ranging from servicing the mortgage to curing delinquent credit cards or paying off medical bills.  What we have discovered is that how a borrower uses discretionary cash from a loan modification relates in a large part to the borrower’s lifestyle, a dimension that requires some specialized analytics to determine.
 
BORROWER HYPOTHETICALS

Let’s examine the life of Lois and Eddie. They are a married couple of 20 years, entrenched in a small community in the Midwest. They have two children in local schools, relationships with many townspeople, have been working for the same employers for more than 15 years and have family and coworkers in the area. They live frugally, borrowing infrequently and make limited use of credit, such as revolving debt.

Lois and Eddie have a much different lifestyle than do Vicky and Dave. Vicky and Dave are a 20-something, newly married couple, who bought a home in a rapidly developing metro-suburban community where they live largely anonymously, barely knowing the names of their neighbors. They have no children, few friends locally and no family nearby. They have been with their respective employers less than five years each and live many miles from work, with no co-workers in the neighborhood. They are more aggressive borrowers with more expensive automobiles bought with borrowed money. They support significantly higher revolving balances and have a penchant to add to their credit portfolio, especially their revolving debt.  

Here’s the question: Will Lois and Eddie respond similarly to Vicky and Dave, given a bout of mortgage distress? Hardly. That’s the point. And, the other point: The terms of the loan modification are highly unlikely to uproot the lifestyle and financial priorities of Lois and Eddie or of Vicky and Dave. Lifestyle differences trump loan modification terms and will do so all day long. Loan modification terms cannot cause Vicky and Dave to become Lois and Eddie anymore more than they can cause Lois and Eddie to behave like Vicky and Dave.  

Lois and Eddie are much more likely to use the discretionary cash from loan modification to save their mortgage, even if that means going severely delinquent or defaulting on credit card debt. For Lois and Eddie, their lifestyle — and their values — demand that they save their home. They derive their identity and their purpose from their community. Losing their home would mean losing membership in the community, which is a big part of what makes Lois and Eddie the people they are.  A foreclosure would cause

Lois and Eddie a great loss-of-face among friends, co-workers and family. Their employers might take note as well.
In contrast, Vicky and Dave are less likely to use discretionary cash to save their mortgage and will only do so if saving their mortgage does little to inconvenience their lifestyle, which involves many other forms of borrowing and consumption. Vicky and Dave derive meaning in their life from factors far removed from their property, their mortgage and their local community. Their constituency lies outside the immediate neighborhood. Home and community mean much less to them, and they would suffer very little embarrassment if they had to leave their house (and its mortgage) behind for other priorities. One of their major priorities is servicing their revolving debt portfolio, which is large and now causing debt servicing challenges.

These couples are really different kinds of people, with different lifestyle priorities. Lois and Eddie would most likely modify their lifestyle extensively (consume less and work extra hours at the mill) to work within the boundaries of a loan modification in order to keep their home. Vicky and Dave are the opposite: They would most likely not modify their lifestyle extensively, nor endure many inconveniences to keep their mortgage. For them, the loan modification must suit their lifestyle, which in this case is based upon aggressive use of revolving credit.

CHANGING AND ANALYZING BEHAVIOR

We believe lifestyle is a function of years of learning, growing and behaving, and these behaviors are very resistant to change as they are closely associated with the borrower’s value system. Lifestyles are not likely to be strongly influenced by external events such as a call from a mortgage collector with a mortgage payment reduction.    

Our research has shown that lifestyle exerts a profound influence on a borrower’s financial behavior in terms of borrowing, paying and prioritizing debt. Which kinds of debt will the borrower save from default and which will be allowed to default and why?  
The linkage between financial behavior and lifestyle has significant implications. In order to effect a strong, persistent change in financial behavior — being successful at a loan modification over a sustained period — often requires significant changes in a person’s lifestyle.  

For example, to get Vicky and Dave to behave like Lois and Eddie (and save their mortgage), Vicky and Dave would have to place much more value on neighborhood and community. This would represent a major change and would result in friction from the other aspects of their lifestyle that they value more.

A BIRD’S EYE VIEW

The Veritas Borrower Segmentation Engine represents a very aggressive form of borrower analysis. It analyzes each borrower in depth and classifies him or her into one of 32 Clusters and one of seven “Meta Clusters” — which are groups of Clusters with common properties. Veritas uses more than 120 attributes about the borrower, the property and the local real estate market to analyze and classify a borrower. Table 1 provides the details of “Meta Cluster” membership, a name and a position for Lois/Eddie and Vicky/Dave.  Table 2 links the “Meta Cluster” name to its position of in the 3D array shown in Exhibit 2.

We have discovered that by using an aggressive form of borrower segmentation (Veritas), we can meaningfully classify borrowers. That classification not only enables us to understand many aspects of a borrower’s current state, but it also enables us to project future states and associated behaviors.  
In effect, borrower segmentation helps us identify and understand a borrower’s lifestyle and estimate financial behavior, including borrowing and paying debts.  Cluster (or Meta Cluster) membership designates the lifestyle and the behavioral tendencies of borrowers in each Cluster are closely associated with the borrower’s future financial behavior, such as succeeding at a loan modification.  

By using the Veritas Borrower Segmentation Engine, we can correctly classify Lois and Eddie vs. Vicky and Dave. Lois and Eddy belong to “Meta G” and Vicky and Dave are assigned to “Meta E.”  

Meta G (Lois and Eddie):  This Meta Cluster is comprised of low income borrowers who are responsible consumers of debt; they very likely own a modest, below median-priced home, view their mortgage as their primary financial obligation and pay limited interest in other forms of debt, such as credit card and retail card.   Property values in their neighborhood have been more stable and not subject to wild speculation and tremendous price increases or decreases.

Meta E (Vicky and Dave):  This Meta Cluster contains borrowers with a much different lifestyle.  Meta E is composed of borrowers who in the recent past have been very aggressive borrowers across a broad spectrum of debt, well beyond mortgage. However, they are beginning to fall behind in their debt servicing and are suffering eroding credit and declining cash to service their debt.  Property values in this Cluster have incurred more depreciation, with many properties now presenting negative equity. This may account for a portion of this borrower’s recent debt servicing distress, since the borrower is no longer able to harvest home equity to support aggressive consumption habits.

Meta Cluster colors translate to Exhibit 2 where all Clusters in a Meta Cluster are colored the same.
Exhibit 2 depicts how the Clusters and Meta Clusters are positioned in a three-dimensional space.  The axes of the 3D array are especially meaningful, since they capture key aspects of borrower and property behavior:  
CLTV: Combined Loan to Value — a measure of the economic value of the property.  Those with CLTV in excess of 100% have negative equity.

Borrower Factor 1:  Borrower Debt Service — a measure of the borrower’s willingness and ability to service big ticket debt, such as a mortgage.
Borrower Factor 2:  Consumption of New Accounts — a measure of how many new accounts of any type (credit card, finance company, auto, mortgage) were acquired within the past 12 months.

The positioning of the Meta Clusters along the three axes shows great dispersion, suggesting many key differences across Clusters and Meta Clusters. Meta E (Vicky and Dave) and Meta G (Lois and Eddie) differ in at least two major ways. Meta E (17, 28, 29) shows an eroding debt service, with more seriously delinquent accounts, as well as an increasing CLTV, with many properties in this Cluster having a CLTV > 100%.  Meta G (Lois and Eddie) is showing a slightly stronger ability to service their debt.  However, Meta G (2, 26) also demonstrates a lower CLTV, with a portion of this population showing a CLTV < 100%, indicating positive equity.  However, resist the temptation to conclude that the differences between Meta E and Meta G can be explained by CLTV effects.  

INFLUENCING FINANCIAL BEHAVIOR

We are discovering that Meta Clusters define lifestyle and lifestyle defines future financial behavior. If this is true, it stands to reason, then that we should be able to predict loan mod redefault behavior from Meta Cluster membership, which we can, as is shown in Exhibit 3 below, which contains statistics from a random sample of 55,000 loans modified in 2009 and tracked through 2011.  

As is revealed by this exhibit, Meta Cluster does aggressively predict among those borrowers who would fail at loan modification (e.g. Meta B, E, (Vicky and Dave) and D) and those who would succeed (e.g. Meta F, G (Lois and Eddie) A and C).  For example, Meta E (redefault:  41%), redefaults at a significantly higher rate than Meta B (redefault: 26%), which is consistent with the lifestyle-based hypothesis representing Vicky and Dave (Meta E) vs. Lois and Eddie (Meta G). Apparently, Lois and Eddie were better equipped than Vicky and Dave to succeed at loan modification. The question is why?

There is the persistent CLTV argument, which goes like this: Since Meta E borrowers own a home with a much higher CLTV than Meta G borrowers (128% v 91%). Then Meta G, being economically rational would be more likely to support and succeed at loan modification. However, as is shown below in Exhibit 4, three of the best performing Meta Clusters (C, A and F) have a CLTV far greater than 100% and near equivalent to the worst performing Meta Clusters (B, E and D).  

If combined loan to value in 2009 exerted a strong effect on loan modification performance from 2009-2011, the trend would be pronounced and obvious. And, it is the opposite.  It is not pronounced and CLTV is not obviously correlated with Meta Cluster loan modification redefault performance. In sum, combined loan to value does not explain much about loan modification redefault behavior.

Exhibit 5 offers another kind of explanation, one based upon the level of debt servicing distress as measured by Meta Cluster. The problem-debt percentage in 2009 correlates quite strongly with Meta Cluster loan modification redefault performance in 2011, save for Meta F, which represents only 2% of all loan modifications (please reference Exhibit 6).  

As is shown in Exhibit 5, the spike in the problem debt percentage in 2009 is strongly associated with a spike in loan modification redefaults in 2011. The problem debt percentage in 2009 for Meta E and B is 26% and 50%, respectively, the two highest rates in the population. In turn, the loan modification redefault rates in 2011 for Meta E and B are also the two highest in the population (41% and 51%, respectively).

Ignoring Meta F, due to its small percentage effect, the trend in problem debt percentage vs. loan modification redefaults shows a clear pattern: Once the problem debt percentage reached critical mass in 2009 — more than  25% — loan modification redefault performance erodes substantially.  

When comparing the relative power of CLTV 2009 vs. problem debt percentage in 2009, it is evident that metrics that reflect borrower behavior (especially debt servicing distress) at a very granular level are more closely associated and possibly causally related to loan modification redefault performance than property economics as indicated by CLTV. This is not to say that property economics does not play a role, but rather such a role has been overstated in many instances.

Exhibit 6 reveals how loan modifications were distributed across Meta Clusters in 2009. Meta F had the smallest percentage, garnering 2% of all available loan modifications.  Ironically, a majority of the loan modifications (56%) were given to borrowers in Meta B and E, who also had the highest problem debt percentage at the time of modification and also redefaulted at the highest rates by 2011. Meta E also contains borrowers Vicky and Dave.

The loan modification allocation process had its inconsistencies, however.  While the data suggests that the weakest borrowers (Meta B, E) were the most likely to receive a loan modification, stronger borrowers were not ignored by the process.  Borrowers in Meta A and G, some of the lowest redefaulting borrowers in the population, also captured a significant minority of the loan modifications: 22% and 3%, respectively.  These borrowers had a significantly lower problem debt percentage and also redefaulted at below average rates (average redefault rate: < 30%).  Lois and Eddie are in Meta G.

The analysis of loan modification allocation by Meta Cluster highlights the inconsistency of loan modification with respect to those borrower attributes that are known to affect loan modification performance. In particular, why would a process allocate a majority of loan modifications (56%) to borrowers in Meta B and E, when these borrowers have some of the highest problem debt levels and are most likely to redefault in excess?
 
APPLYING DISCRETIONARY CASH

Besides problem debt, is there another borrower-centric element that can further explain why some borrowers redefault at higher rates. Exhibit 7 analyzes how borrowers applied the discretionary cash from loan modification.  

In Exhibit 7, those Meta Clusters, B and E, which redefaulted at the highest rates (51% and 41%, respectively), were organized into one group:  high redefault rate group. The remainder (Meta Clusters A, C, D, F and G), which redefaulted at far lower rates were organized into the second group: low redefault rate group.  (Redefault rates of 30%, 34%, 37%, 11% and 26%, respectively.) 

Vicky and Dave are in the high redefault group while Lois and Eddie are in the low redefault group.
The ratio in 2009 indicates how borrowers from each group were most likely to apply their discretionary cash. The metric compares the rate at which the borrowers in the group allowed their seriously delinquent revolving debt to roll to a more serious level of delinquency, that was more than 60 days past due.   None of the accounts rolling to into greater delinquency were in bankruptcy or foreclosure. Nor were they settled in lieu of default.

If the ratio is less than 100%, then the group is dedicating its resources to reducing the proportion of delinquent revolving debt that rolls to a more serious delinquency level; if the ratio is greater than 100%, then the group for one reason or another is allowing a greater proportion its seriously delinquent revolving debt to roll to a higher level of delinquency.

It is clear that the borrowers in the high redefault rate group applied their discretionary cash to their distressed revolving debt, significantly reducing the proportion that rolled to a more serious level of delinquency. In 2009, 38% of the high redefault rate group’s revolving debt rolled to debt that was more than 60 days overdue. However, by 2011, only 16% of this group’s revolving accounts were rolling to more than 60 days over due, or 42% of the 2009 rate of 38% (16%/38% = 42%).  

The low redefault group behaved much differently. By 2011, they were letting their revolving debt role to more than 60 days overdue at a much higher rate. In 2009, this group allowed 14% of its delinquent revolving debt to roll over into more than 60 days overdue. However, by 2011, this group was allowing more than 25% of its seriously delinquent revolving debt roll to a status of more than 60 days overdue or 175% of the 2009 rate (25%/14% = 175%).  

This is a dramatic change in behavior and indicates that the low redefault rate group was highly likely to apply their discretionary cash to their mortgage loan.  They did this, even though they would incur a larger number of revolving defaults, which would negatively impact their credit score, possibly denying them borrowing opportunities, especially in the near future.

BORROWER-CENTRIC OUTCOMES

This behavorial look at the borrower in a Clustering concept captures a borrower’s lifestyle and shows how that lifestyle is intimately associated with present and future borrower behavior. Knowledge of borrower lifestyle enables an analyst to estimate loan modification redefault behavior.  

However, the ability to gain insight doesn’t not stop there. The analyst could accurately deduce not only who would redefault and who would not, but provide excellent insight into why a borrower would redefault.  It enabled the analyst, for example, to determine which borrowers were most likely to use the discretionary cash from loan modification to improve mortgage debt service and which borrowers would apply that cash for other things such as paying down credit cards.

The analysis challenges existing methods of mortgage analysis, which are decidedly nonborrower centric, focusing on such elements as loan type, loan modification terms and property economics such as negative equity issues. It is not that methods of mortgage analysis are incorrect, wrong or useless; rather they fail to extract and use a very granular understanding of the borrower, which can be exceptionally valuable in borrower intensive arenas such as loan modification.  

One of the failures of the traditional analysis methodology was the failure of CLTV to explain borrower loan modification behavior.  Since CLTV-based interpretations of future loan behavior are a dominant influence in the industry, the introduction of a complementary and potentially contradictory analysis methodology is at the least interesting and potentially profound. More borrower-centric metrics were needed to explain behavior, such as problem debt percentage, which morphed into Meta Clusters. Clusters were then linked directly to loan modification redefault behavior, which was shown to be essentially independent of CLTV effects.

The findings reveal how complex analyzing and understanding borrower behavior can be, especially distressed borrower behavior. Moreover, it reveals how such well-meaning distressed loan treatments often lead to numerous unintended consequences. For example, with a specific kind of borrower, loan modification causes the mortgage lender/investor to incur a reduction in the cash value of the loan.  The lender/investor incurs this loss in the hope of mitigating future losses.  I doubt if the lender/investor incurs this additional expense in order to improve the borrower’s bankcard portfolio. Yet, that is exactly what has happened in some instances.

There is a growing trend in which a majority of the distressed mortgage borrowers are more prone to use the discretionary cash from loan modification to cure non-mortgage debt servicing problems. This means, in reality, there are many more couples like Vicky and Dave and far fewer like Lois and Eddie.

FUTURE LENDING RISKS

The state of the borrower appears to be evolving, in particular his or her willingness to service big ticket debt such as a mortgage. The morality of owing money is gradually being replaced by a mindset of optimizing how much is owed and to whom in order to maximize one’s lifestyle concerns. Such concerns are often distinct and separate from the morality of borrowing and owing money.  

If this is true (or becoming true), then lenders have a great deal to think about as they lend money, assess risk and predict the cash flow from financial assets, such as mortgage loans. If these less-moral forms of lifestyle are becoming commonplace, it is important to recognize how deeply embedded and resistant to change such lifestyles really are.  

It is imperative for the lender and investor to know when they are investing in a consumer financial asset that is ultimately dependent upon the borrower’s morality and lifestyle in order for the asset to perform and the investor and lender to earn a return. Minimizing or perhaps directing one’s exposure to this kind of borrower may be a valid consideration in the “new normal” in which we are likely to reside for some time to come. 

Most Popular Articles

Latest Articles

Lower mortgage rates attracting more homebuyers 

An often misguided premise I see on social media is that lower mortgage rates are doing nothing for housing demand. That’s ok — very few people are looking at the data without an agenda. However, the point of this tracker is to show you evidence that lower rates have already changed housing data. So, let’s […]

3d rendering of a row of luxury townhouses along a street

Log In

Forgot Password?

Don't have an account? Please