Get Started
Articles Tagged with: RS Edge

Prepayment Spikes in Ida’s Wake – What to Expect

It is, of course, impossible to view the human suffering wrought by Hurricane Ida without being reminded of Hurricane Katrina’s impact 16 years ago. Fortunately, the levees are holding and Ida’s toll appears likely to be less severe. It is nevertheless worth taking a look at what happened to mortgages in the wake of New Orleans’s last major catastrophic weather event as it is reasonable to assume that prepayments could follow a similar pattern (though likely in a more muted way).

Following Katrina, prepayment speeds for pools of mortgages located entirely in Louisiana spiked between November 2005 and June 2006. As the following graph shows, prepayment speeds on Louisiana properties (the black curve) remained elevated relative to properties nationally (the blue curve) until the end of 2006. 

Comparing S-curves of Louisiana loans (the black curve in the chart below) versus all loans (the green curve) during the spike period (Nov. 2005 to Jun. 2006) reveals speeds ranging from 10 to 20 CPR faster across all refinance incentives. The figure below depicts an S-curve for non-spec 100% Louisiana pools and all non-spec pools with a weighted average loan age of 7 to 60 months during the period indicated.

The impact of Katrina on Louisiana prepayments becomes even more apparent when we consider speeds prior to the storm. As the S-curves below show, non-specified 100% Louisiana pools (the black curve) actually paid slightly slower than all non-spec pools between November 2003 and October 2005.

As we pointed out in June, a significant majority of prepayments caused by natural disaster events are likely to be voluntary, as opposed to the result of default as one might expect. This is because mortgages on homes that are fully indemnified against these perils are likely to be prepaid using insurance proceeds. This dynamic is reflected in the charts below, which show elevated voluntary prepayment rates running considerably higher than the delinquency spike in the wake of Katrina. We are able to isolate voluntary prepayment activity by looking at the GSE Loan Level Historical Performance datasets that include detailed credit information. This enables us to confirm that the prepay spike is largely driven by voluntary prepayments. Consequently, recent covid-era policy changes that may reduce the incidence of delinquent loan buyouts from MBS are unlikely to affect the dynamics underlying the prepayment behavior described above.

RiskSpan’s Edge Platform enables users to identify Louisiana-based loans and pools by drilling down into cohort details. The example below returns over $1 billion in Louisiana-only pools and $70 billion in Louisiana loans as of the August 2021 factor month.


Edge also allows users to structure more specified queries to identify the exposure of any portfolio or portfolio subset. Edge, in fact, can be used to examine any loan characteristic to generate S-curves, aging curves, and time series.  Contact us to learn more.



EDGE: QM vs Non-QM Prepayments

Prepayment speeds for qualified mortgages (QM loans) have anecdotally been faster than non-QM loans. For various reasons, the data necessary to analyze interest rate incentive response has not been readily available for these categories of mortgages.

In order to facilitate the generation of traditional refinancing curves (S-curves) over the last year, we have normalized data to improve the differentiation of QM versus non-QM loans within non-agency securities.

Additionally, we isolated the population to remove prepay impact from loan balance and seasoning.

The analysis below was performed on securitized loans with 9 to 36 months of seasoning and an original balance between 200k and 500k. S-curves were generated for observation periods from January 2016 through July 2021.

Results are shown in the table and chart below.

 


 

For this analysis, refinance incentive was calculated as the difference between mortgage note rate and the 6-week lagged Freddie Mac primary mortgage market survey (PMMS) rate. Non-QM borrowers would not be able to easily refi into a conventional mortgage. We further analyzed the data by examining prepayments speeds for QM and non-QM loans at different level of SATO. SATO, the spread at origination, is calculated as the difference between mortgage note rate and the prevailing PMMS rate at time of loan’s origination.

 

Using empirical data maintained by RiskSpan, it can be seen the refinance response for QM loans remains significantly faster than Non-QM loans.

Using Edge, RiskSpan’s data analytics platform, we can examine any loan characteristic and generate S-curves, aging curves, and time series. If you are interested in performing historical analysis on securitized loan data, please contact us for a free demonstration.


EDGE: Extended Delinquencies in Loan Balance Stories

In June, we highlighted Fannie Mae’s and Freddie Mac’s new “expanded delinquency” states. The Enterprises are now reporting delinquency states from 1 to 24 months to better account for loans that are seriously delinquent and not repurchased under the extended timeframe for repurchase of delinquent loans announced in 2020.

This new data reveals a strong correlation between loan balance and “chronically delinquent” loans. In the graph below, we chart loan balance on the x-axis and 180+Day delinquency on the y-axis, for 2017-18 production 30yr 3.5s through 4.5 “generic” borrowers.[1]

As the graph shows, within a given coupon, loans with larger original balances also tended to have higher “chronic delinquencies.

The graph above also illustrates a clear correlation between higher chronic delinquencies and higher coupons. This phenomenon is most likely due to SATO. While each of these queries excluded low-FICO, high-LTV, and NY loans, the 2017-18 30yr 3.5 cohort was mostly at-the-money origination, whereas 4.0s and 4.5s had an average SATO of 30bp and 67bp respectively. The higher SATO indicates a residual credit quality issue. As one would expect, and we demonstrated in our June analysis, lower-credit-quality loans tend also to have higher chronic delinquencies.

The first effect – higher chronic delinquencies among larger loans within a coupon – is more challenging to understand. We posit that this effect is likely due to survivor bias. The large refi wave over the last 18 months has factored-down higher-balance cohorts significantly more than lower-balance cohorts.

Higher-credit-quality borrowers tend to refinance more readily than lower-credit-quality borrowers, and because the larger-loan-balance cohorts have seen higher total prepayments, these same cohorts are left with a larger residue of lower-quality credits. The impact of natural credit migration (which is observed in all cohorts) tends to leave behind a larger proportion of credit-impaired borrowers in faster-paying cohorts versus the slower-paying, lower-loan-balance cohorts.

The higher chronic delinquencies in larger-loan-balance cohorts may ultimately lead to higher buyouts, depending on the resolution path taken. As loan balance decreases, the lower balance cohorts will have reduced risk to these potential buyouts, leaving them better protected to any uptick in involuntary speeds.


Contact us if you are interested in seeing variations on this theme. Using Edge, we can examine any loan characteristic and generate a S-curve, aging curve, or time series.


[1] We filtered for borrowers with LTV<=80, FICO>=700, and ex-NY. We chose 2017-18 production to analyze, to give sufficient time for loans to go chronically delinquent. We see a similar relationship in 2019 production, see RiskSpan for details.


EDGE: Extended Delinquencies in FNMA and FHLMC Loans

In June, the market got its first look at Fannie Mae and Freddie Mac “expanded delinquency” states. The Enterprises are now reporting delinquency states out to 24 months to better account for loans that are seriously delinquent and not repurchased under the extended timeframe for repurchase of delinquent loans announced in 2020. In this short post, we analyze those pipelines and what they could mean for buyouts in certain spec pool stories. 

First, we look at the extended pipeline for some recent non-spec cohorts. The table below summarizes some major 30yr cohorts and their months delinquent. We aggregate the delinquencies that are more than 6 months delinquent[1] for ease of exposition. 

Recent-vintage GSE loans with higher coupons show a higher level of “chronically delinquent” loans, similar to the trends we see in GNMA loans. 

Digging deeper, we filtered for loans with FICO scores below 680. Chronically delinquent loan buckets in this cohort are marginally more prevalent relative to non-spec borrowers. Not unexpectedly, this suggests a credit component to these delinquencies.

Finally, we filtered for loans with high LTVs at origination. The chronically delinquent buckets are lower than the low FICO sector but still present an overhang of potential GSE repurchases in spec pools.

It remains to be seen whether some of these borrowers will be able to resume their original payments —  in which case they can remain in the pool with a forbearance payment due at payoff — or if the loans will be repurchased by the GSEs at 24 months delinquent for modification or other workout. If the higher delinquencies lead to the second outcome, the market could see an uptick in involuntary speeds on some spec pool categories in the next 6-12 months.


Contact us if you are interested in seeing variations on this theme. Using Edge, we can examine any loan characteristic and generate a S-curve, aging curve, or time series.


[1] The individual delinquency states are available for each bucket, contact us for details.


Data & Machine Learning Workshop Series

RiskSpan’s Edge Platform is supported by a dynamic team of professionals who live and breathe mortgage and structured finance data. They know firsthand the challenges this type of data presents and are always experimenting with new approaches for extracting maximum value from it.

Join us for a complimentary series of virtual workshops where RiskSpan professionals share what we’re learning about applying machine learning and other innovative techniques to data that asset managers, broker-dealers and mortgage bankers care about.

Catch up with these previously recorded workshops


Measuring and Visualizing Feature Impact & Machine Learning Model Materiality

RiskSpan CIO Suhrud Dagli demonstrates in greater detail how machine learning can be used in input data validations, to measure feature impact, and to visualize how multiple features interact with each other.

Structured Data Extraction from Images Using Google Document AI

RiskSpan Director Steven Sun shares a procedural approach to tackling the difficulties of efficiently extracting structured data from images, scanned documents, and handwritten documents using Google’s latest Document AI Solution.

Pattern Recognition in Time Series Data

Traders and investors rely on time series patterns generated by asset performance to inform and guide their trading and asset allocation decisions. Economists take advantage of analogous patterns in macroeconomic and market data to forecast recessions and other market events. But you need to be able to spot these patterns in order to use them.

Advanced Forecasting Using Hierarchical Models

Traditional statistical models apply a single set of coefficients by pooling a large dataset or for specific cohorts. Hierarchical models learn from feature behavior across dimensions or timeframes. This informative workshop applies hierarchical models to a variety of mortgage and structured finance use cases.

Quality Control with Anomaly Detection (Part I)

Outliers and anomalies refer to various types of occurrences in a time series. Spike of value, shift in level or volatility or a change in seasonal pattern are common examples.  RiskSpan Co-Founder & CIO Suhrud Dagli is joined by Martin Kindler, a market risk practitioner who has spent decades dealing with outliers.

Quality Control with Anomaly Detection (Part 2)

Suhrud Dagli presents Part 2 of this workshop, which dove into mortgage loan QC and introduce coding examples and approaches for avoiding false negatives using open-source Python algorithms in the Anomaly Detection Toolkit (ADTK).


Mortgage DQs by MSA: Non-Agency Performance Chart of the Month

This month we take a closer look at geographical differences in loan performance in the non-agency space. The chart below looks at the 60+ DPD Rate for the 5 Best and Worst performing MSAs (and the overall average). A couple of things to note:

  • The pandemic seems to have simply amplified performance differences that were already apparent pre-covid. The worst performing MSAs were showing mostly above-average delinquency rates before last year’s disruption.
  • Florida was especially hard-hit. Three of the five worst-performing MSAs are in Florida. Not surprisingly, these MSAs rely heavily on the tourism industry.
  • New York jumped from being about average to being one of the worst-performing MSAs in the wake of the pandemic. This is not surprising considering how seriously the city bore the pandemic’s brunt.
  • Tech hubs show strong performance. All our best performers are strong in the Tech industry—Austin’s the new Bay Area, right?
Contact Us

Anomaly Detection and Quality Control

In our most recent workshop on Anomaly Detection and Quality Control (Part I), we discussed how clean market data is an integral part of producing accurate market risk results. As incorrect and inconsistent market data is so prevalent in the industry, it is not surprising that the U.S. spends over $3 trillion on processes to identify and correct market data.

In taking a step back, it is worth noting what drives accurate market risk analytics. Clearly, having accurate portfolio holdings with correct terms and conditions for over-the-counter trades is central to calculating consistent risk measures that are scaled to the market value of the portfolio. The use of well-tested and integrated industry-standard pricing models is another key factor in producing reliable analytics. In comparison to the two categories above, clean, and consistent market data are the largest contributors that could lead to poor market risk analytics. The key driving factor behind detecting and correcting/transforming market data is risk and portfolio managers expectation that risk results are accurate at the start of the business day with no need to perform any time-consuming re-runs during the day to correct issues found.  

Broadly defined, market data is defined as any data that is used as input to the re-valuation models. This includes equity prices, interest rates, credit spreads. FX rates, volatility surfaces, etc.

Market data needs to be:

  • Complete – no true gaps when looking back historically.
  • Accurate
  • Consistent – data must be viewed across other data points to determine its accuracy (e.g., interest rates across tenor buckets, volatilities across volatility surface)

Anomaly types can be broken down into four major categories:

  • Spikes
  • Stale data
  • Missing data
  • Inconsistencies

Here are three example of “bad” market data:

Credit Spreads

The following chart depicts day-over-day changes in credit spreads for the 10-year consumer cyclical time series, returned from an external vendor. The changes indicate a significant spike on 12/3 that caused big swings, up and down, across multiple rating buckets​. Without an adjustment to this data, key risk measures would show significant jumps, up and down, depending on the dollar value of positions on two consecutive days​.

Swaption Volatilities

Market data also includes volatilities, which drive delta and possible hedging. The following chart shows implied swaption volatilities for different maturities of swaptions and their underlying swaps. The following chart shows implied swaption volatilities for different maturity of swaption and underlying swap​. Note the spikes in 7×10 and 10×10 swaptions. The chart also highlights inconsistencies between different tenors and maturities.

Equity Implied Volatilities

The 146 and 148 strikes in the table below reflect inconsistent vol data, as often occurs around expiration.

The detection of market data inconsistencies needs to be an automated process with multiple approaches targeted for specific types of market data. The detection models need to evolve over time as added information is gathered with the goal of reducing false negatives to a manageable level. Once the models detect the anomalies, the next step is to automate the transformation of the market data (e.g., backfill, interpolate, use prior day value). Together with the transformation, transparency must be recorded such that it is known what values were either changed or populated if not available. This should be shared with clients which could lead to alternative transformations or model detection routines.

Detector types typically fall into the following categories:

  • Extreme Studentized Deviate (ESD): finds outliers in a single data series (helpful for extreme cases.)
  • Level Shift: detects change in level by comparing means of two sliding time windows (useful for local outliers.)
  • Local Outliers: detects spikes in near values.
  • Seasonal Detector: detects seasonal patterns and anomalies (used for contract expirations and other events.)
  • Volatility Shift: detects shift of volatility by tracking changes in standard deviation.

 

On Wednesday, May 19th, we will present a follow-up workshop focusing on:

  • Coding examples
    • Application of outlier detection and pipelines
    • PCA
  • Specific loan use cases
    • Loan performance
    • Entity correction
  • Novelty Detection
    • Anomalies are not always “bad”
    • Market monitoring models

You can register for this complimentary workshop here.


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.


Three Principles for Effectively Monitoring Machine Learning Models

The recent proliferation in machine learning models in banking and structured finance is becoming impossible to ignore. Rarely does a week pass without a client approaching us to discuss the development or validation (or both) of a model that leverages at least one machine learning technique. RiskSpan’s own model development team has also been swept up in the trend – deep learning techniques have featured prominently in developing the past several versions of our in-house residential mortgage prepayment model.  

Machine learning’s rise in popularity is attributable to multiple underlying trends: 

  1. Quantity and complexity of data. Nowadays, firms store every conceivable type of data relating to their activities and clients – and frequently supplement this with data from any number of third-party providers. The increasing dimensionality of data available to modelers makes traditional statistical variable selection more difficult. The tradeoff between a model’s complexity and the rules adapted in variable selection can be hard to balance. An advantage of ML approaches is that they can handle multi-dimensional data more efficiently. ML frameworks are good at identifying trends and patterns – without the need for human intervention. 
  2. Better learning algorithms. Because ML algorithms learn to make more accurate projections as new data is introduced to the framework (assuming there is no data bias in the new data) model features based on newly introduced data are more likely to resemble features created using model training data.  
  3. Cheap computation costsNew techniques, such as XGBoost, are designed to be memory efficient. It introduces an innovated system design that helps in reducing the computation cost. 
  4. Proliferation breeds proliferation. As the number of machine learning packages in various programming tools increases, it facilitates implementation and promotes further ML model development. 

Addressing Monitoring Challenges 

Notwithstanding these advances, machine learning models are by no means easy to build and maintain. Feature engineering and parameter tuning procedures are time consuming. And once a ML model has been put into production, monitoring activities must be implemented to detect anomalies to make sure the model works as expected (just like with any other model). According to the OCC 2011-12 supervisory guidance on the model risk management, ongoing monitoring is essential to evaluate whether changes in products, exposures, activities, clients, or market conditions necessitate adjustment, redevelopment, or replacement of the model and to verify that any extension of the model beyond its original scope is valid. While monitoring ML models resembles monitoring conventional statistical models in many respects, the following activities take on particular importance with ML model monitoring: 

  1. Review the underlying business problem. Defining the business problem is the first step in developing any ML model. This should be carefully articulated in the list of business requirements that the ML model is supposed to follow. Any shift in the underlying business problem will likely create drift in the training data and, as a result, new data coming to the model may no longer be relevant to the original business problem. The ML model becomes degraded and the new process of feature engineering and parameter tuning needs to be considered to remediate the impact. This review should be conducted whenever the underlying problem or requirements change. 
  2.  Review of data stability (model input). In the real world, even if the underlying business problem is unchanged, there might be shifts in the predicting data caused by changing borrower behaviors, changes in product offerings, or any other unexpected market drift. Any of these things could result in the ML model receiving data that it has not been trained on. Model developers should measure the data population stability between the training dataset and the predicting dataset. If there is evidence of the data having shifted, model recalibration should be considered. This assessment should be done when the model user identifies significant shift in the model’s performance or when a new testing dataset is introduced to the ML model. Where data segmentation has been used in the model development process, this assessment should be performed at the individual segment level, as well. 
  3. Review of performance metrics (model output). Performance metrics quantify how well an ML model is trained to explain the data. Performance metrics should fit the model’s type. For instance, the developer of a binary classification model could use Kolmogorov-Smirnov (KS) table, receiver operating characteristic (ROC) curve, and area under the curve (AUC) to measure the model’s overall rank order ability and its performance at different cutoffs. Any shift (upward or downward) in performance metrics between a new dataset and the training dataset should raise a flag in monitoring activity. All material shifts need to be reviewed by the model developer to determine their cause. Such assessments should be conducted on an annual basis or whenever new data is available. 

Like all models, ML models are only as good as the data they are fed. But ML models are particularly susceptible to data shifts because their processing components are less transparent. Taking these steps to ensure they are learning based on valid and consistent data are essential to managing a functional inventory of ML models. 


EDGE: New Forbearance Data in Agency MBS

Over the course of 2020 and into early 2021, the mortgage market has seen significant changes driven by the COVID pandemic. Novel programs, ranging from foreclosure moratoriums to payment deferrals and forbearance of those payments, have changed the near-term landscape of the market.

In the past three months, Fannie Mae and Freddie Mac have released several new loan-level credit statistics to address these novel developments. Some of these new fields are directly related to forbearance granted during the pandemic, while others address credit performance more broadly.

We summarize these new fields in the table below. These fields are all available in the Edge Platform for users to query on.

The data on delinquencies and forbearance plans covers March 2021 only, which we summarize below, first by cohort and then by major servicer. Edge users can generate other cuts using these new filters or by running the “Expanded Output” for the March 2021 factor date.

In the first table, we show loan-level delinquency for each “Assistance Plan.” Approximately 3.5% of the outstanding GSE universe is in some kind of Assistance Plan.

In the following table, we summarize delinquency by coupon and vintage for 30yr TBA-eligible pools. Similar to delinquencies in GNMA, recent-vintage 3.5% and 4.5% carry the largest delinquency load.

Many of the loans that are 90-day and 120+-day delinquent also carry a payment forbearance. Edge users can simultaneously filter for 90+-day delinquency and forbearance status to quantify the amount of seriously delinquent loans that also carry a forbearance versus loans with no workout plan.[2]  Finally, we summarize delinquencies by servicer. Notably, Lakeview and Wells leads major servicers with 3.5% and 3.3% of their loans 120+-day delinquent, respectively. Similar to the cohort analysis above, many of these seriously delinquent loans are also in forbearance. A summary is available on request.

In addition to delinquency, the Enterprises provide other novel performance data, including a loan’s total payment deferral amount. The GSEs started providing this data in December, and we now have sufficient data to start to observing prepayment behavior for different levels of deferral amounts. Not surprisingly, loans with a payment deferral prepay more slowly than loans with no deferral, after controlling for age, loan balance, LTV, and FICO. When fully in the money, loans with a deferral paid 10-13 CPR slower than comparable loans.

Next, we separate loans by the amount of payment deferral they have. After grouping loans by their percentage deferral amount, we observe that deferral amount produces a non-linear response to prepayment behavior, holding other borrower attributes constant.

Loans with deferral amounts less than 2% of their UPB showed almost no prepayment protection when deep in-the-money.[3] Loans between 2% and 4% deferral offered 10-15 CPR protection, and loans with 4-6% of UPB in deferral offered a 40 CPR slowdown.

Note that as deferral amount increases, the data points with lower refi incentive disappear. Since deferral data has existed for only the past few months, when 30yr primary rates were in a tight range near 2.75%, that implies that higher-deferral loans also have higher note rates. In this analysis, we filtered for loans that were no older than 48 months, meaning that loans with the biggest slowdown were typically 2017-2018 vintage 3.5s through 4.5s.

Many of the loans with P&I deferral are also in a forbearance plan. Once in forbearance, these large deferrals may act to limit refinancings, as interest does not accrue on the forborne amount. Refinancing would require this amount to be repaid and rolled into the new loan amount, thus increasing the amount on which the borrower is incurring interest charges. A significantly lower interest rate may make refinancing advantageous to the borrower anyway, but the extra interest on the previously forborne amount will be a drag on the refi savings.

Deferral and forbearance rates vary widely from servicer to servicer. For example, about a third of seriously delinquent loans serviced by New Residential and Matrix had no forbearance plan, whereas more than 95% of such loans serviced by Quicken loans were in a forbearance plan. This matters because loans without a forbearance plan may ultimately be more subject to repurchase and modification, leading to a rise in involuntary prepayments on this subset of loans.

As the economy recovers and borrowers increasingly resolve deferred payments, tracking behavior due to forbearance and other workout programs will help investors better estimate prepayment risk, both due to slower prepays as well as possible future upticks in buyouts of delinquent loans.


Contact us if you are interested in seeing variations on this theme. Using Edge, we can examine any loan characteristic and generate a S-curve, aging curve, or time series.




[1] A link to the Deferral Amount announcement can be found here, and a link to the Forbearance and Delinquency announcement can be found here. Freddie Mac offers a helpful FAQ here on the programs.

[2] Contact RiskSpan for details on how to run this query.

[3] For context, a payment deferral of 2% represents roughly 5 months of missed P&I payments on a 3% 30yr mortgage.


Get Started