Linkedin    Twitter   Facebook

Get Started
Log In

Linkedin

Blog Archives
Home / Blog Archives

Why Mortgage Climate Risk is Not Just for Coastal Investors

When it comes to climate concerns for the housing market, sea level rise and its impacts on coastal communities often get top billing. But this article in yesterday’s New York Times highlights one example of far-reaching impacts in places you might not suspect.

Chicago, built on a swamp and virtually surrounded by Lake Michigan, can tie its whole existence as a city to its control and management of water. But as the Times article explains, management of that water is becoming increasingly difficult as various dynamics related to climate change are creating increasingly large and unpredictable fluctuations in the level of the lake (higher highs and lower lows). These dynamics are threatening the city with more frequency and severe flooding.

The Times article connects water management issues to housing issues in two ways: the increasing frequency of basement flooding caused by sewer overflow and the battering buildings are taking from increased storm surge off the lake. Residents face increasing costs to mitigate their exposure and fear the potentially negative impact on home prices. As one resident puts it, “If you report [basement flooding] to the city, and word gets out, people fear it’s going to devalue their home.”

These concerns — increasing peril exposure and decreasing valuations — echo fears expressed in a growing number of seaside communities and offer further evidence that mortgage investors cannot bank on escaping climate risk merely by avoiding the coasts. Portfolios everywhere are going to need to begin incorporating climate risk into their analytics.

Hurricane Season a Double-Whammy for Mortgage Prepayments

As hurricane (and wildfire) season ramps up, don’t sleep on the increase in prepayment speeds after a natural disaster event. The increase in delinquencies might get top billing, but prepays also increase after events—especially for homes that were fully insured against the risk they experienced. For a mortgage servicer with concentrated geographic exposure to the event area, this can be a double-whammy impacting their balance sheet—delinquencies increase servicing advances, prepays rolling loans off the book. Hurricane Katrina loan performance is a classic example of this dynamic.

Hurrican-Season-a-Double-Whammy-for-Mortgage

Non-Agency Delinquencies Fall Again – Still Room for Improvement

Serious delinquencies among non-Agency residential mortgages continue marching downward during the first half of 2021 but remain elevated relative to their pre-pandemic levels.

Our analysis of more than two million loans held in private-label mortgage-backed securities found that the percentage of loans at least 60 days past due fell again in May across vintages and FICO bands. While performance differences across FICO bands were largely as expected, comparing pre-crisis vintages with mortgages originated after 2009 revealed some interesting distinctions.

The chart below plots serious delinquency rates (60+ DPD) by FICO band for post-2009 vintages. Not surprisingly, these rates begin trending upward in May and June of 2020 (two months after the economic effects of the pandemic began to be felt) with the most significant spikes coming in July and August – approaching 20 percent at the low end of the credit box and less than 5 percent among prime borrowers.

Since last August’s peak, serious delinquency rates have fallen most precipitously (nearly 8 percentage points) in the 620 – 680 FICO bucket, compared with a 5-percentage point decline in the 680 – 740 bucket and a 4 percentage point drop in the sub-620 bucket. Delinquency rates have come down the least among prime (FICO > 740) mortgages (just over 2 percentage points) but, having never cracked 5 percent, these loans also had the shortest distance to go.

Serious delinquency rates remain above January 2020 levels across all four credit buckets – approximately 7 percentage points higher in the two sub-680 FICO buckets, compared with the 680 – 740 bucket (5 percentage points higher than in January 2020) and over-740 bucket (2 percentage points higher).

So-called “legacy” vintages (consisting of mortgage originated before the 2008-2009 crisis) reflect a somewhat different performance profile, though they follow a similar pattern.

The following chart plots serious delinquency rates by FICO band for these older vintages. Probably because these rates were starting from a relatively elevated point in January 2020, their pandemic-related spike were somewhat less pronounced, particularly in the low-FICO buckets. These vintages also appear to have felt the spike about a month earlier than did the newer issue loans.

Serious delinquency rates among these “legacy” loans are considerably closer to their pre-pandemic levels than are their new-issue counterparts. This is especially true in the sub-prime buckets. Serious delinquencies in the sub-620 FICO bucket actually were 3 percentage points lower last month than they were in January 2020 (and nearly 5 percentage points lower than their peak in July 2020). These differences are less pronounced in the higher-FICO buckets but are still there.

Comparing the two graphs reveals that the pandemic had the effect of causing new-issue low-FICO loans to perform similarly to legacy low-FICO loans, while a significant gap remains between the new-issue prime buckets and their high-FICO pre-2009 counterparts. This is not surprising given the tightening that underwriting standards (beyond credit score) underwent after 2009.

Interested in cutting non-Agency performance across any of several dozen loan-level characteristics? Contact us for a quick, no-pressure demo.

Leveraging ML to Enhance the Model Calibration Process

Last month, we outlined an approach to continuous model monitoring and discussed how practitioners can leverage the results of that monitoring for advanced analytics and enhanced end-user reporting. In this post, we apply this idea to enhanced model calibration.

Continuous model monitoring is a key part of a modern model governance regime. But testing performance as part of the continuous monitoring process has value that extends beyond immediate governance needs. Using machine learning and other advanced analytics, testing results can also be further explored to gain a deeper understanding of model error lurking within sub-spaces of the population.

Below we describe how we leverage automated model back-testing results (using our machine learning platform, Edge Studio) to streamline the calibration process for our own residential mortgage prepayment model.

The Problem:

MBS prepayment models, RiskSpan’s included, often provide a number of tuning knobs to tweak model results. These knobs impact the various components of the S-curve function, including refi sensitivity, turnover lever, elbow shift, and burnout factor.

The knob tuning and calibration process is typically messy and iterative. It usually involves somewhat-subjectively selecting certain sub-populations to calibrate, running back-testing to see where and how the model is off, and then tweaking knobs and rerunning the back-test to see the impacts. The modeler may need to iterate through a series of different knob selections and groupings to figure out which combination best fits the data. This is manually intensive work and can take a lot of time.

As part of our continuous model monitoring process, we had already automated the process of generating back-test results and merging them with actual performance history. But we wanted to explore ways of taking this one step further to help automate the tuning process — rerunning the automated back-testing using all the various permutations of potential knobs, but without all the manual labor.

The solution applies machine learning techniques to run a series of back-tests on MBS pools and automatically solve for the set of tuners that best aligns model outputs with actual results.

We break the problem into two parts:

  1. Find Cohorts: Cluster pools into groups that exhibit similar key pool characteristics and model error (so they would need the same tuners).

TRAINING DATA: Back-testing results for our universe of pools with no model tuning knobs applied

  1. Solve for Tuners: Minimize back-testing error by optimizing knob settings.

TRAINING DATA: Back-testing results for our universe of pools under a variety of permutations of potential tuning knobs (Refi x Turnover)

  1. Tuning knobs validation: Take optimized tuning knobs for each cluster and rerun pools to confirm that the selected permutation in fact returns the lowest model errors.

Part 1: Find Cohorts

We define model error as the ratio of the average modeled SMM to the average actual SMM. We compute this using back-testing results and then use a hierarchical clustering algorithm to cluster the data based on model error across various key pool characteristics.

Hierarchical clustering is a general family of clustering algorithms that build nested clusters by either merging or splitting observations successively. The hierarchy of clusters is represented as a tree (or dendrogram). The root of the tree is the root cluster that contains all samples, while the leaves represent clusters with only one sample. [1]

Agglomerative clustering is an implementation of hierarchical clustering that takes the bottom-up approach (merging approach). Each observation starts in its own cluster, and clusters are then successively merged together. There are multiple linkage criteria that could be chosen from. We have used Ward linkage criteria.

Ward linkage strategy minimizes the sum of squared differences within all clusters. It is a variance-minimizing approach.[2]

Part 2: Solving for Tuners

Here our training data is expanded to be a set of back-test results to include multiple results for each pool under different permutations of tuning knobs.  

Process to Optimize the Tuners for Each Cluster

Training Data: Rerun the back-test with permutations of REFI and TURNOVER tunings, covering all reasonably possible combinations of tuners.

  1. These permutations of tuning results are fed to a multi-output regressor, which trains the machine learning model to understand the interaction between each tuning parameter and the model as a fitting step.
    • Model Error and Pool Features are used as Independent Variables
    • Gradient Tree Boosting/Gradient Boosted Decision Trees (GBDT)* methods are used to find the optimized tuning parameters for each cluster of pools derived from the clustering step
    • Two dependent variables — Refi Tuner and Turnover Tuner – are used
    • Separate models are estimated for each cluster
  2. We solve for the optimal tuning parameters by running the resulting model with a model error ratio of 1 (no error) and the weighted average cluster features.

* Gradient Tree Boosting/Gradient Boosted Decision Trees (GBDT) is a machine learning technique for regression and classification problems, which produces a prediction model in the form of an ensemble of weak prediction models, typically decision trees. When a decision tree is a weak learner, the resulting algorithm is called gradient boosted trees, which usually outperforms random forest. It builds the model in a stage-wise fashion like other boosting methods do, and it generalizes them by allowing optimization of arbitrary differentiable loss function. [3]

*We used scikit-learn’s GBDT implementation to optimize and solve for best Refi and Turnover tuner. [4]

Results

The resultant suggested knobs show promise in improving model fit over our back-test period. Below are the results for two of the clusters using the knobs that suggested by the process. To further expand the results, we plan to cross-validate on out-of-time sample data as it comes in.

Conclusion

These advanced analytics show promise in their ability to help streamline the model calibration and tuning process by removing many of the time-consuming and subjective components from the process altogether. Once a process like this is established for one model, applying it to new populations and time periods becomes more straightforward. This analysis can be further extended in a number of ways. One in particular we’re excited about is the use of ensemble models—or a ‘model of models’ approach. We will continue to tinker with this approach as we calibrate our own models and keep you apprised on what we learn.

Too Many Documentation Types? A Data-Driven Approach to Consolidating Them

The sheer volume of different names assigned to various documentation types in the non-agency space has really gotten out of hand, especially in the last few years. As of February 2021, an active loan in the CoreLogic RMBS universe could have any of over 250 unique documentation type names, with little or no standardization from issuer to issuer. Even within a single issuer, things get complicated when every possible permutation of the same basic documentation level gets assigned its own type. One issuer in the database has 63 unique documentation names!

In order for investors to be able to understand and quantify their exposure, we need a way of consolidating and mapping all these different documentation types to a simpler, standard nomenclature. Various industry reports attempt to group all the different documentation levels into meaningful categories. But these classifications often fail to capture important distinctions in delinquency performance among different documentation levels.

There is a better way. Taking some of the consolidated group names from the various industry papers and rating agency papers as a starting point, we took another pass focusing on two main elements:

  • The delinquency performance of the group. We focused on the 60-DPD rate while also considering other drivers of loan performance (e.g., DTI, FICO, and LTV) and their correlation to the various doc type groups.
  • The size of the sub-segment. We ensured our resulting groupings were large enough to be meaningful.

What follows is how we thought about it and ultimately landed where we did. These mappings are not set in stone and will likely need to undergo revisions as 1) new documentation types are generated, and 2) additional performance data and feedback from clients on what they consider most important become available. Releasing these mappings into RiskSpan’s Edge Platform will then make it easier for users to track performance.

Data Used

We take a snapshot of all loans outstanding in non-agency RMBS issued after 2013, as of the February 2021 activity period. The data comes from CoreLogic and we exclude loans in seasoned or reperforming deals. We also exclude loans whose documentation type is not reported, some 14 percent of the population.

Approach

We are seeking to create sub-groups that generally conform to the high-level groups on which the industry seems to be converging while also identifying subdivisions with meaningfully different delinquency performance. We will rely on these designations as we re-estimate our credit model.

Steps in the process:

  1. Start with high-level groupings based on how the documentation type is currently named.
    • Full Documentation: Any name referencing ‘Agency,’ ‘Agency AUS,’ or similar.
    • Bank Statements: Any name including the term “Bank Statement[s].”
    • Investor/DSCR: Any name indicating that the underwriting relied on net cash flows to the secured property.
    • Alternative Documentation: A wide-ranging group consolidating many different types, including: asset qualifier, SISA/SIVA/NINA, CPA letters, etc.
    • Other: Any name that does not easily classify into one of the groups above, such as Foreign National Income, and any indecipherable names.

Chart

  1. We subdivided the Alternative Documentation group by some of the meaningfully sized natural groupings of the names:
    • Asset Depletion or Asset Qualifier
    • CPA and P&L statements
    • Salaried/Wage Earner: Includes anything with W2 tax return
    • Tax Returns or 1099s: Includes anything with ‘1099’ or ‘Tax Return, but not ‘W2.’
    • Alt Doc: Anything that remained, included items like ‘VIVA, ‘SISA,’ ‘NINA,’ ‘Streamlined,’ ‘WVOE,’ and ‘Alt Doc.’
  1. From there we sought to identify any sub-groups that perform differently (as measured by 60-DPD%).
    • Bank Statement: We evaluated a subdivision by the number of statements provided (less than 12 months, 12 months, and greater than 12 months). However, these distinctions did not significantly impact delinquency performance. (Also, very few loans fell into the under 12 months group.) Distinguishing ‘Business Bank Statement’ loans from the general ‘Bank Statements’ category, however, did yield meaningful performance differences.

High Level

    • Alternative Documentation: This group required the most iteration. We initially focused our attention on documentation types that included terms like ‘streamlined’ or ‘fast.’ This, however, did not reveal any meaningful performance differences relative to other low doc loans. We also looked at this group by issuer, hypothesizing that some programs might perform better than others. The jury is still out on this analysis and we continue to track it. The following subdivisions yielded meaningful differences:
      • Limited Documentation: This group includes any names including the terms ‘reduced,’ ‘limited,’ ‘streamlined,’ and ‘alt doc.’ This group performed substantially better than the next group.
      • No Doc/Stated: Not surprisingly, these were the worst performers in the ‘Alt Doc’ universe. The types included here are a throwback to the run-up to the housing crisis. ‘NINA,’ ‘SISA,’ ‘No Doc,’ and ‘Stated’ all make a reappearance in this group.
      • Loans with some variation of ‘WVOE’ (written verification of employment) showed very strong performance, so much so that we created an entirely separate group for them.
  • Full Documentation: Within the variations of ‘Full Documentation’ was a whole sub-group with qualifying terms attached. Examples include ‘Full Doc 12 Months’ or ‘Full w/ Asset Assist.’ These full-doc-with-qualification loans were associated with higher delinquency rates. The sub-groupings reflect this reality:
      • Full Documentation: Most of the straightforward types indicating full documentation, including anything with ‘Agency/AUS.’
      • Full with Qualifications (‘Full w/ Qual’): Everything including the term ‘Full’ followed by some sort of qualifier.
  • Investor/DSCR: The sub-groups here either were not big enough or did not demonstrate sufficient performance difference.
  • Other: Even though it’s a small group, we broke out all the ‘Foreign National’ documentation types into a separate group to conform with other industry reporting.

High Level

Among the challenges of this sort of analysis is that the combinations to explore are virtually limitless. Perhaps not surprisingly, most of the potential groupings we considered did not make it into our final mapping. Some of the cuts we are still looking at include loan purpose with respect to some of the alternative documentation types.

We continue to evaluate these and other options. We can all agree that 250 documentation types is way too many. But in order to be meaningful, the process of consolidation cannot be haphazard. Fortunately, the tools for turning sub-grouping into a truly data-driven process are available. We just need to use them.

Value Beyond Validation: The Future of Automated Continuous Model Monitoring Has Arrived

Imagine the peace of mind that would accompany being able to hand an existing model over to the validators with complete confidence in how the outcomes analysis will turn out. Now imagine being able to do this using a fully automated process.

The industry is closer to this than you might think.

The evolution of ongoing model monitoring away from something that happens only periodically (or, worse, only at validation time) and toward a more continuous process has been underway for some time. Now, thanks to automation and advanced process design, this evolutionary process has reached an inflection point. We stand today at the threshold of a future where:

  • Manual, painful processes to generate testing results for validation are a thing of the past;
  • Models are continuously monitored for fit, and end users are empowered with the tools to fully grasp model strengths and weaknesses;
  • Modeling and MRM experts leverage machine learning to dive more deeply into the model’s underlying data, and;
  • Emerging trends and issues are identified early enough to be addressed before they have time to significantly hamper model performance.

Sound too good to be true? Beginning with its own internally developed prepayment and credit models, RiskSpan data scientists are laying out a framework for automated, ongoing performance monitoring that has the potential to transform behavioral modeling (and model validation) across the industry.

The framework involves model owners working collaboratively with model validators to create recurring processes for running previously agreed-upon tests continuously and receiving the results automatically. Testing outcomes continuously increases confidence in their reliability. Testing them automatically frees up high-cost modeling and validation resources to spend more time evaluating results and running additional, deeper analyses.

The Process:

Irrespective of the regulator, back-testing, benchmarking, and sensitivity analysis are the three pillars of model outcomes analysis. Automating the data and analytical processes that underlie these three elements is required to get to a fully comprehensive automated ongoing monitoring scheme.

In order to be useful, the process must stage testing results in a central database that can:

  • Automatically generate charts, tables, and statistical tests to populate validation reports;
  • Support dashboard reporting that allows model owners, users and validators to explore test results, and;
  • Feed advanced analytics and machine learning platforms capable of 1) helping with automated model calibration, and 2) identifying model weaknesses and blind spots (as we did with a GSE here).

Perhaps not surprisingly, achieving the back-end economies of a fully automated continuous monitoring and reporting regime requires an upfront investment of resources. This investment takes the form of time from model developers and owners as well as (potentially) some capital investment in technology necessary to host and manage the storage of results and output reports.

A good rule of thumb for estimating these upfront costs is between 2 and 3 times the cost of a single annual model test performed on an ad-hoc, manual basis. Consequently, the automation process can generally be expected to pay for itself (in time savings alone) over 2 to 3 cycles of performance testing. But the benefits of automated, continuous model monitoring go far beyond time savings. They invariably result in better models.

Output Applications

Continuous model monitoring produces benefits that extend well beyond satisfying model governance requirements. Indeed, automated monitoring has significantly informed the development process for RiskSpan’s own, internally developed credit and prepayment models – specifically in helping to identify sub-populations where model fit is a problem.

Continuous monitoring also makes it possible to quickly assess the value of newly available data elements. For example, when the GSEs start releasing data on mortgages with property inspection waivers (PIWs) (as opposed to traditional appraisals) we can immediately combine that data element with the results of our automated back-testing to determine whether the PIW information can help predict model error from those results. PIW currently appears to have value in predicting our production model error, and so the PIW feature is now slated to be added to a future version of our model. Having an automated framework in place accelerates this process while also enabling us to proceed with confidence that we are only adding variables that improve model performance.

The continuous monitoring results can also be used to develop helpful dashboard reports. These provide model owners and users with deeper insights into a model’s strengths and weaknesses and can be an important tool in model tuning. They can also be shared with model validators, thus facilitating that process as well.

The dashboard below is designed to give our model developers and users a better sense of where model error is greatest. Sub-populations with the highest model error are deep red. This makes it easy for model developers to visualize that the model does not perform well when FICO and LTV data are missing, which happens often in the non-agency space. The model developers now know that they need to adjust their modeling approach when these key data elements are not available.

The dashboard also makes it easy to spot performance disparities by shelf, for example, and can be used as the basis for applying prepayment multipliers to certain shelves in order to align results with actual experience.

Continuous model monitoring is fast becoming a regulatory expectation and an increasingly vital component of model governance. But the benefits of continuous performance monitoring go far beyond satisfying auditors and regulators. Machine learning and other advanced analytics are also proving to be invaluable tools for better understanding model error within sub-spaces of the population.

Watch this space for a forthcoming post and webinar explaining how RiskSpan leverages its automated model back-testing results and machine learning platform, Edge Studio, to streamline the calibration process for its internally developed residential mortgage prepayment model.

Flood Insurance Changes: What Mortgage Investors Need to Know

Major changes are coming to FEMA’s National Flood Insurance Program on April 1st, 2021, the impacts of which will reverberate throughout real estate, mortgage, and structured finance markets in a variety of ways. 

For years, the way the NFIP has managed flood insurance in the United States has been the subject of intense scrutiny and debate. Compounding the underlying moral hazard issues raised by the fact that taxpayers are subsidizing homeowners who knowingly move into flood-prone areas is the reality that the insurance premiums paid by these homeowners collectively are nowhere near sufficient to cover the actual risks faced by properties in existing flood zones. 

Climate change is only exacerbating the gap between risk and premiums. According to research released this week by First Street Foundation, the true economic risk is 3.7 times higher than the level at which the NFIP is currently pricing flood insurance. And premiums would need to increase by 7 times to cover the expected economic risk in 2051. 

A New York Times article this week addresses some of the challenges (political and otherwise) a sudden increase in flood insurance premiums would create. These include existing homeowners no longer being able to afford the higher monthly payments as well as a potential drop in property values in high-risk areas as the cost of appropriately priced flood insurance is priced in. These risks are also of concern to mortgage investors who obviously have little interest in seeing sudden declines in the value of properties that secure the mortgages they own. 

Notwithstanding these risks, the NFIP recognizes that the disparity between true risk and actual premiums cannot continue to go unaddressed. The resulting adjustment to the way in which the NFIP will calculate premiums – called Risk Rating 2.0 – will reflect a policy of phasing out subsidies (wherein lower-risk dwellings absorb the cost of those in the highest-risk areas) and tying premiums to the actual flood risk of a given structure. 

Phase-In 

The specific changes to be announced on April 1st will go into effect on October 1st, 2021. But the resulting premium increases will not happen all at once. Annual limits currently restrict how fast premiums can increase for primary residences, ranging from 5%-18% per year. (Non-primary residences have a cap of 25%). FEMA has not provided much guidance on how these caps will apply under Risk Rating 2.0 other than to say that “all properties will be on a glide path to actuarial rates.” The caps, however, are statutory and would require an act of Congress to change. And Members of Congress have shown reluctance in the past to saddle their constituents with premium spikes. 

Phasing in premium increases helps address the issue of affordability for current homeowners. This is equally important to investors who hold these existing homeowners’ mortgages. It does not however, address the specter of significant property value declines because the sale of the home has historically caused the new, fully priced premium to take effect for the next homeowner. It has been suggested that FEMA could blunt this problem by tying insurance premiums to properties rather than to homeowners. This would enable the annual limits on price increases to remain in effect even if the house is sold. 

Flood Zones & Premiums 

Despite a widely held belief that flood zone maps are out of date and that climate change is hastening the need to redraw them, Risk Rating 2.0 will reportedly apply only to homes located in floodplains as currently defined. Premium calculations, however, will focus on the geographical and structural features of a particular home, including foundation type and replacement cost, rather than on a property’s location within a flood zone.  

The Congressional Research Service’s paper detailing Risk Rating 2.0 acknowledges that premiums are likely to go up for many properties that are currently benefiting from subsidies. The paper emphasizes that it is not in FEMA’s authority to provide affordability programs and that this is a job for Congress as they consider changes to the NFIP. 

“FEMA does not currently have the authority to implement an affordability program, nor does FEMA’s current rate structure provide the funding required to support an affordability program. However, affordability provisions were included in the three bills which were introduced in the 116th Congress for long-term reauthorization of the NFIP: the National Flood Insurance Program Reauthorization Act of 2019 (H.R. 3167), and the National Flood Insurance Program Reauthorization and Reform Act of 2019 (S. 2187) and its companion bill in the House, H.R. 3872. As Congress considers a long-term reauthorization of the NFIP, a central question may be who should bear the costs of floodplain occupancy in the future and how to address the concerns of constituents facing increases in flood insurance premiums.” 

Implications for Homeowners and Mortgage Investors 

FEMA is clearly signaling that NFIP premium increases are coming. Any increases to insurance premiums will impact the value of affected homes in much the same way as rising interest rates. Both drive prices down by increasing monthly payments and thus reducing the purchasing power of would-be buyers. The difference, however, is that while interest rates affect the entire housing market, this change will be felt most acutely by owners of properties in FEMA’s Special Flood Hazard Areas that require insurance. The severity of these impacts will clearly be related to the magnitude of the premium increases, whether increase caps will be applied to properties as well as owners, and the manner in which these premiums get baked into sales prices. 

Mortgage risk holders need to be ready to assess their exposure to these flood zone properties and the areas that see the biggest rate jumps. The simplest way to do this is through HPI scenarios based on a consistent view of the ‘affordability’ of the house — i.e., by adjusting the maximum mortgage payment for a property downward to compensate for the premium increase and then solving for the drag on home price.

Get in touch with us for a no-obligation discussion on how to measure the impact of these forthcoming changes on your portfolio. We’d be interested in hearing your insights as well. 

Overcoming Data Limitations (and Inertia) to Factor Climate into Credit Risk Modeling

With each passing year, it is becoming increasingly clear to mortgage credit investors that climate change is emerging as a non-trivial risk factor that must be accounted for. Questions around how precisely to account for this risk, however, and who should ultimately bear it, remain unanswered. 

Current market dynamics further complicate these questions. Late last year, Politico published this special report laying out the issues surrounding climate risk as it relates to mortgage finance. Even though almost everyone agrees that underinsured natural disaster risk is a problem, the Politico report outlines several forces that make it difficult for anyone to do anything about it. The massive undertaking of bringing old flood zone maps up to date is just one example. As Politico puts it: 

The result, many current and former federal housing officials acknowledge, is a peculiar kind of stasis — a crisis that everyone sees coming but no one feels empowered to prevent, even as banks and investors grow far savvier about assessing climate risk. 

At some point, however, we will reach a tipping point – perhaps a particularly devastating event (or series of events) triggering significant losses. As homeowners, the GSEs, and other mortgage credit investors point fingers at one another (and inevitably at the federal government) a major policy update will become necessary to identify who ultimately bears the brunt of mispriced climate risk in the market. Once quantified and properly assigned, the GSEs will price in climate risk in the same way they bake in other contributors to credit risk — via higher guarantee fees. For non-GSE (and CRT) loans, losses will continue to be borne by whoever holds the credit risk.  

Recognizing that such an event may not be far off, the GSEs, their regulator, and everyone else with credit exposure are beginning to appreciate the importance of understanding the impact of climate events on mortgage performance. This is not easily inferred from the historical data record, however. And those assessing risk need to make informed assumptions about how historically observed impacts will change in the future. 

The first step in constructing these assumptions is to compile a robust historical dataset. To this end, RIskSpan began exploring the impact of certain hurricanes a few years ago. This initial analysis revealed a significant impact on short-term mortgage delinquency rates (not surprisingly), but less of an impact on default rates. In other words, affected borrowers encountered hardship but ultimately recovered. 

This research is preliminary, however, and more data will be necessary to build scenario assumptions as climate events become more severe and widespread. As more data covering more events—including wildfires—becomes available, RiskSpan is engaged in ongoing research to tease out the impact each of these events has on mortgage performance.  

It goes without saying that climate scenario assumptions need to be grounded in reality to be useful to credit investors. Because time-series data relationships are not always detectable using conventional means, especially when data is sparse, we are beginning to see promise in leveraging various machine learning techniques to this end. We believe this historical, machine-learning-based research will provide the backbone for an approach that merges historical effects of events with inputs about the increasing frequency and severity of these events as they become better understood and more quantifiable. 

Precise forecasting of severe climate events by zip code in any given year is not here yet. But an increasingly reliable framework for gauging the likely impact of these events on mortgage performance is on the horizon.  

The NRI: An Emerging Tool for Quantifying Climate Risk in Mortgage Credit

Climate change is affecting investment across virtually every sector in a growing number of mostly secondary ways. Its impact on mortgage credit investors, however, is beginning to be felt more directly.

Mortgage credit investors are investors in housing. Because housing is subject to climate risk and borrowers whose houses are destroyed by natural disasters are unlikely to continue paying their mortgages, credit investors have a vested interest in quantifying the risk of these disasters.

To this end, RiskSpan is engaged in leveraging the National Risk Index (NRI) to assess the natural disaster and climate risk exposure of mortgage portfolios.

This post introduces the NRI data in the context of mortgage portfolio analysis (loans or mortgage-backed securities), including what the data contain and key considerations when putting together an analysis. A future post will outline an approach for integrating this data into a framework for scenario analysis that combines this data with traditional mortgage credit models.

The National Risk Index

The National Risk Index (NRI) was released in October 2020 through a collaboration led by FEMA. It provides a wealth of new geographically specific data on natural hazard risks across the country. The index and its underlying data were designed to help local governments and emergency planners to better understand these risks and to plan and prepare for the future.

The NRI provides information on both the frequency and severity of natural risk events. The level of detailed underlying data it provides is astounding. The NRI focuses on 18 natural risks (discussed below) and provides detailed underlying components for each. The severity of an event is broken out by damage to buildings, agriculture, and loss of life. This breakdown lets us focus on the severity of events relative to buildings. While the definition of building here includes all types of real estate—houses, commercial, rental, etc.—having the breakdown provides an extra level of granularity to help inform our analysis of mortgages.

The key fields that provide important information for a mortgage portfolio analysis are bulleted below. The NRI provides these data points for each of the 18 natural hazards and each geography they include in their analysis.

  • Annualized Event Frequency
  • Exposure to Buildings: Total dollar amount of exposed buildings
  • Historical Loss Ratio for Buildings (Bayesian methods to derive this estimate, such that every geography is covered for its relevant risks)
  • Expected Annual Loss for Buildings
  • Population estimates (helpful for geography weighting)

Grouping Natural Disaster Risks for Mortgage Analysis

The NRI data covers 18 natural hazards, which pose varying degrees of risk to housing. We have found the framework below to be helpful when considering which risks to include in an analysis. We group the 18 risks along two axes:

1) The extent to which an event is impacted by climate change, and

2) An event’s potential to completely destroy a home.

Earthquakes, for example, have significant destructive potential, but climate change is not a major contributor to earthquakes. Conversely, heat waves and droughts wrought by climate change generally do not pose significant risk to housing structures.

When assessing climate risk, RiskSpan typically focuses on the five natural hazard risks in the top right quadrant below.

Immediate Event Risk versus Cumulative Event Risk

Two related but distinct risks inform climate risk analysis.

  1. Immediate Event Analysis: The risk of mortgage delinquency and default resulting directly from a natural disaster event. A home severely damaged or destroyed by a hurricane, for example.  
  2. Cumulative Event Risk: Less direct than immediate event risk, this is the risk of widespread home price declines across an entire area communities because of increasing natural hazard risk brought on by climate change. These secondary effects include: 
    • Heightened homebuyer awareness or perception of increasing natural hazard risk,
    • Property insurance premium increases or areas becoming ‘self-insured,’ 
    • Government policy impacts (e.g., potential flood zone remapping), and 
    • Potential policy changes related to insurance from key players in the mortgage market (i.e., Fannie Mae, Freddie Mac, FHFA, etc.). 

NRI data provides an indication of the probability of immediate event occurrence and its historic severity in terms of property losses. We can also empirically observe historical mortgage performance in the wake of previous natural disaster events. Data covering several hurricane and wildfire events are available.

Cumulative event risk is less observable. A few academic papers attempt to tease out these impacts, but the risk of broader home price declines typically needs to be incorporated into a risk assessment framework through transparent scenario overlays. Examples of such scenarios include home price declines of as much as 20% in newly flood-exposed areas of South Florida. There is also research suggesting that there are often long term impacts to consumer credit following a natural disaster.  

Geography Normalization

Linking to the NRI is simple when detailed loan pool geographic data are available. Analysts can merge by census tract or county code. Census tract is the more geographically granular measure and provides a more detailed analysis.

For many capital markets participants, however, that level of geographic specific detail is not available. At best, an investor may have a 5-digit or 3-digit zip code. Zip codes do not directly match to a given county or census tract and can potentially span across those distinctions.

There is no perfect way to perform the data link when zip code is the only available geographic marker. We take an approach that leverages the other data on housing stock by census tract to weight mortgage portfolio data when multiple census tracts map to a zip code.

Other Data Limitations

The loss information available represents a simple historical average loss rate given an event. But hurricanes (and hurricane seasons) are not all created equal. The same is true of other natural disasters. Relying on averages may work over long time horizons but could significantly underpredict or overpredict loss in a particular year. Further, the frequency of events is rising so that what used to be considered 100 year event may be closer to a 10 or 20 year event. Lacking data about what losses might look like under extreme scenarios makes modeling such events problematic.

The data also make it difficult to take correlation into account. Hurricanes and coastal flooding are independent events in the dataset but are obviously highly correlated with one another. The impact of a large storm on one geographic area is likely to be correlated with that of nearby areas (such as when a hurricane makes its way up the Eastern Seaboard).

The workarounds for these limitations have limitations of their own. But one solution involves designing transparent assumptions and scenarios related to the probability, severity, and correlation of stress events. We can model outlier events by assuming that losses for a particular peril follow a normal distribution with set standard deviations. Other assumptions can be made about correlations between perils and geographies. Using these assumptions, stress scenarios can be derived by picking a particular percentile along the loss distribution.

A Promising New Credit Analysis Tool for Mortgages

Notwithstanding its limitations, the new NRI data is a rich source of information that can be leveraged to help augment credit risk analysis of mortgage and mortgage-backed security portfolios. The data holds great promise as a starting point (and perhaps more) for risk teams starting to put together climate risk and other ESG analysis frameworks.

Logo

Company

Products

Security & Compliance

Linkedin

SOC2 Type II Certified

Get Started
Log in

Linkedin   

risktech2024