Get Started
Category: Article

RS Edge: WALA Ramps for Non-Bank Servicers

In 2019, the non-bank servicing sector continued to grow faster than traditional bank-servicers. As a group, non-bank servicers now represent nearly half of the agency MBS market, with outsized representation in newer-production mortgages. Their aggressive refinancing has driven speeds on in-the-money mortgages to post-crisis highs, and we believe this behavior will continue into 2020.  

But within the non-bank sector, prepayment behavior varies widely. In this short post, we measure the fastest non-bank servicers against their cohorts and against the wider market. 

We used the Edge platform to generate WALA ramps for the top 25 non-bank servicers for 30yr “generic” mortgages.¹ In the first graph, we show WALA ramps for bank-serviced and non-bankserviced loans that were 75-125bp in the money over the last calendar year. At the peak, non-bank servicers outstripped bank servicers by roughly 8 CPR. 

In the next chart, we break out performance for the two fastest non-bank servicers: United Shore and Provident Funding.² United Shore clocked in at blazing 83 CPR for the 7-8 WALA bucket with Provident printing in the high 70s. 

Age-Bucket-vs-CPR

Switching to SMMthe right way to examine such fast speedswe see that loans serviced by United Shore paid at 13.7 SMM, more than twice the unscheduled principal per month than the cohort of non-bank servicers in months 7 and 8. 

  Age-Bucket-vs-SMM

In closing, we note that newer vintage Freddie Mac Supers consistently contain more United Shore and Provident product than similarly aged Fannie Mae Majors. Together, United Shore and Provident account for 14-18% of newerproduction Freddie Supers, such as FR SD8016, SD8005, SD8001, and SD8006, but only 4-6% of Fannie Majors, such as FN MA3774 or MA3745. Most of the fast-payer Freddie Supers are 3s and 3.5s and may not show fast speeds at current rates, but in a 25-50bp rally we may see separation between Fannie and Freddie TBA speeds. As a consequence, Freddie Supers may have worse convexity than similar vintage Fannie Majors. 

If you are interested in seeing variations on this theme, contact us. Using RS Edge, we can examine any loan characteristic and generate a S-curve, WALA curve, or time series. [/vc_column_text][/vc_column][/vc_row][vc_row][vc_column][vc_empty_space][vc_empty_space][startapp_separator border_width=”1″ opacity=”25″ animation=””][/vc_column][/vc_row][vc_row][vc_column][vc_column_text]¹For a loan to be included, it had to be securitized into a deliverable 30yr Fannie or Freddie pool and have a loan balance greater than $225,000, FICO > 700, LTV <= 80, and not in NY state. All analysis was done at loan level.

²New Residential and Home Point Financial receive an honorable mention for fast speeds. Their speeds showed more response for loans 50-100bp in the money but started to converge to average non-bank speeds when 75-125bp in the money. See RiskSpan for details.


RS Edge for Loans & Structured Products: A Data Driven Approach to Pre-Trade and Pricing  

The non-agency residential-mortgage-backed-securities (RMBS) market has high expectations for increased volume in 2020. Driven largely by expected changes to the qualified mortgage (QM) patch, private-label securities (PLS) issuers and investors are preparing for a 2020 surge. The tight underwriting standards of the post-crisis era are loosening and will continue to loosen if debt-to-income restrictions are lifted with changes to the QM patch 

PLS programs can differ greatly. It’s increasingly important to understand the risks inherent in each underlying poolAt the same time, investment opportunities with substantial yield are becoming harder to find without developing a deep understanding of the riskier components of the capital structureA structured approach to pre-trade and portfolio analytics can help mitigate some of these challenges. Using a data-driven approach, portfolio managers can gain confidence in the positions they take and make data influenced pricing decisions 

Industry best practice for pre-trade analysis is to employ a holistic approach to RMBS. To do this, portfolio managers must combine analysis of loan collateral, historical data for similar cohorts of loans (within previous deals), and scenariofor projected performance. The foundation of this approach is:  

  • Historical data can ground assumptions about projected performance 
  • A consistent approach from deal to deal will illuminate shifting risks from shifting collateral 
  • Scenario analysis will inform risk assessment and investment decision  

Analytical Framework 

RiskSpan’s modeling and analytics expert, Janet Jozwik, suggests a framework for analyzing a new RMBS deal with analysis of 3 main components:  deal collateral, historical performance, and scenario forecasting. Combined, these three components give portfolio managers a present, past, and future view into the deal.  

Present: Deal Collateral Analysis 

Deal collateral analysis consists of: 1) a deep dive into the characteristics of the collateral underlying the deal itself, and 2) a comparison of the collateral characteristics of the deal being analyzed to similar deals. A comparison to recently issued deals can highlight shifts in underlying collateral risk within a particular shelf or across issuers.  

Below, RiskSpan’s RS Edge provides the portfolio manager with a dashboard highlighting key collateral characteristics that may influence deal performance. 

Example 1Deal Profile Stratification 

deal-compare-in-rs-edge

Example 2Deal Comparative Analysis 

Past: Historical Performance Analysis 

Historical analysis informs users of a deal’s potential performance under different scenarios by looking at how similar loan cohorts from prior deals have performedJozwik recommends analyzing historical trends both from the recent past and frohistorical stress vintages to give a sense for what the expected performance of the deal will be, and what the worst-case performance would be under stress scenarios. 

Recent Trend Analysis:  Portfolio managers can understand expected performance by looking at how similar deals have been performing over the prior 2 to 3 years. There are a significant number of recently issued PLS that can be tracked to understand recent prepayment and default trends in the market. While the performance of these recent deals doesn’t definitively determine expectations for a new deal (as things can change, such as rate environment), it provides one data point to help ground data-driven analyses. This approach allows users to capitalize on the knowledge gained from prior market trends.  

Historical Vintage Proxy Analysis:  Portfolio managers can understand stressed performance of the deal by looking at performance of similar loans from vintages that experienced the stress environment of the housing crisisThough potentially cumbersome to execute, this approach leverages the rich set of historical performance data available in the mortgage space 

For a new RMBS Dealportfolio managers can review the distribution of key features, such as FICO, LTV, and documentation typeThey can calculate performance metrics, such as cumulative loss and default rates, from a wide set of historical performance data on RMBS, cut by vintage. When pulling these historical numbers, portfolio managers can adjust the population of loans to better align with the distribution of key loan features in the deal they are analyzing. So, they can get a view into how a similar loans pool originated in historical vintages, like 2007, performed. There are certainly underwriting changes that have occurred in the post-crisis era that would likely make this analysis ultraconservative. These ‘proxy cohorts’ from historical vintages can provide an alternative insight into what could happen in a worst-case scenario.  

Future: Forecasting Scenario Analysis 

Forecasting analysis should come in two flavors. First, very straightforward scenarios that are explicitly transparent about assumptions for CPR, CDR, and severity. These assumptions-based scenarios can be informed with outputs from the Historical Performance Analysis above.  

Second, forecasting analysis can leverage statistical models that consider both loan features and macroeconomic inputs. Scenarios can be built around macroeconomic inputs to the model to better understand how collateral and bond performance will change with changing economic conditions.  Macroeconomic inputs, such as mortgage rates and home prices, can be specified to create particular scenario runs. 

How RiskSpan Can Help 

Pulling the required data and models together is typically a burdenRiskSpan’s RS Edge has solved these issues and now offers one integrated solution for:  

  • Historical Data: Loan-level performance and collateral data on historical and pre-issue RMBS deals 
  • Predictive Models: Credit and Prepayment models for non-agency collateral types 
  • Deal Cashflow Engine: Intex is the leading source for an RMBS deal cashflow library 

There is a rich source of data, models, and analytics that can support decision making in the RMBS market. The challenge for a portfolio manager is piecing these often-disparate pieces of information together to a cohesive analysis that can provide a consistent view from deal to dealFurther, there is a massive amount of historical data in the mortgage space, containing a vast wealth of insight to help inform investment decisions. However, these datasets are notoriously unwieldy. Users of RS Edge cut through the complications of large, disparate datasets for clear, informative analysis, without the need for custom-built technology or analysts with advanced coding skills.


Introducing: RS Edge for Loans and Structured Products

RiskSpan Introduces RS Edge for Loans and Structured Products  

RiskSpan, the leading mortgage data and analytics provider, is excited to announce the release of RS Edge for Loans and Structured Products. 

RS Edge is the next generation of RiskSpan’s data, modeling, and analytics platform that manages portfolio risk and delivers powerful analysis for loans and structured products.  Users can derive insights from historical trends and powerful predictive forecasts under a range of economic scenarios on our cloud-native solution. RS Edge streamlines analysis by bringing together key industry data and integrations with leading 3rd party vendors. 

An on-demand team of data scientists, quants, and technologists with fixed-income portfolio expertise support the integration, calibration, and operation across all RS Edge modules 

RMBS Analytics in Action 

RiskSpan has developed a holistic approach to RMBS analysis that combines loan collateral, historical, and scenario analysis with deal comparison tools to more accurately predict future performance. Asset managers can define an acceptable level of risk and ground pricing decisions with data-driven analysis. This approach illuminates risk from shifting collateral and provides investors with confidence in their positions. 

Loan Analytics in Action 

Whole loan asset managers and investors use RiskSpan’s Loan Analytics to enhance and automate partnerships with Non-Qualified Mortgage originators and servicers. The product enhances the on-boarding, pricing analytics, forecasting, and storage of loan data for historical trend analytics. RS Edge forecasting analytics support ratesheet validation and loan pricing 

About RiskSpan 

RiskSpan provides innovative technology and services to the financial services industry. Our mission is to eliminate inefficiencies in loans and structured finance markets to improve investors’ bottom line through incremental cost savings, improved return on investment, and mitigated risk.  

RiskSpan is holding a webinar on November 6 to show how RS Edge pulls together past, present, and future for insights into new RMBS deals. Click below to register.


EDGE: Revisiting WALA-ramps on FNMA Majors

In the past few months, recent-vintage FNMA Major pools have shown significant acceleration in prepay speeds, significantly impacting TBA prices and dollar rolls. In our August report, we showed a progression of ever faster WALA ramps on FNMA Major pools1. In this installment, we update that behavior using data from Edge, the online prepayment graphing tool.

We start with a population of recent FNMA Majors and generate WALA ramps at loan level, to capture the precise behavior of the WALA ramp. In the first chart, we show loans from Majors that are 75-125bp in the money, approximately TBA 4s, over three different time periods:

  1. August 2018 to July 2019 (“baseline”)
  2. August-September 2019
  3. October 2019

In October, aggregate speeds on Majors hit a new high of 60 CPR for loans in the 9-10 WALA range. More troubling: the tail of the WALA ramp moved higher by roughly 5 CPR. This acceleration impacts carry in the 12mo+ seasoning range and is a potential negative for valuations in the TBA sector.



Graph: Speeds on loans from FN Major pools, holding refi incentive 75-125bp over three different periods.

In the next graph, we use Edge to isolate loans in Major pools that are 25-75bp in the money (approximately 3.5s). Similar to 4s, the progression in the aging curve shows the same story: a faster tail for loans 10+ months seasoned.

WALA Curve and Prepayment Speeds Graph

Graph: Speeds on loans from FN Major pools, holding refinancing incentive 25-75bp over three different periods.

We next look at the change in prepayment speeds from the Aug-Sep period to October and attribute that change to the origination channel. On average, FNMA Major pools are 50:50 Retail origination versus TPO, and we break down the speed contribution into these two groups. In the analysis below, we look at the speed change in each WALA bucket.

For Major 3.5s, the TPO loans accelerated more than the Retail origination loans. But in Major 4.0s, the speeds increased almost equally across each bucket.

fn3.5-major-graphfn4.0-major-graph

In summary, the WALA ramp for TPO is more sensitive than Retail loans when refinancing incentive is small. But when loans are far enough in the money the increase in the WALA ramps are evenly distributed across origination channel.

We continue to monitor the ever-accelerating speeds on FNMA Majors and Freddie Giants, but the trend is clear – the fastest, cheapest to deliver TBA continues to be faster for longer. This makes the ongoing analysis of prepays, whether specified pools or non-spec deliverables, more important that it has been in previous rate cycles.

If you interested in seeing variations on this theme, contact us. Using Edge, we can examine any loan characteristic and generate a S-curve, WALA ramp, or time series.

1See RiskSpan for a similar analysis on newer WALA multi-lender Giants


Navigating the Impact of ASU 2016-13 on AFS Securities

In Collaboration With Our Partners at Grant Thornton

Navigating the impact of ASU 2016-13 on the impairment of AFS debt securities

When the Financial Accounting Standards Board (FASB) issued Accounting Standards Update (ASU) 2016-13, Financial Instruments – Credit Losses, in June of 2016, most of the headlines regarding the ASU focused on its introduction of Subtopic 326-20, commonly referred to as the Current Expected Credit Losses (or, “CECL”) framework.  The CECL framework requires entities to measure lifetime expected credit losses on all financial instruments measured at amortized cost – financial assets like loans receivable and held-to-maturity debt securities.  The focus on the CECL framework was understandable – it represents a sea change in the accounting for a significant class of assets for many entities, particularly lending institutions.

However, ASU 2016-13 affected the accounting for credit losses on other financial instruments as well, such as debt securities held as available-for-sale (or “AFS”).  Below, we will discuss how ASU 2016-13 changed the accounting for credit losses on AFS debt securities.

AFS Framework prior to adopting ASU 2016-13:  OTTI

Prior to an entity’s adoption of ASU 2016-13, the guidance concerning impairment of AFS debt securities is found in Subtopic 320-10, particularly in paragraphs 320-10-35-18 through 35-34, and is known as the Other-Than-Temporary Impairment (or “OTTI”) framework.

Generally, AFS debt securities are carried on the balance sheet at fair value, and changes in the fair value of AFS debt securities are recognized outside of earnings as a component of Other Comprehensive Income (OCI). However, if an AFS debt security’s fair value is less than its amortized cost – that is, the AFS debt security is impaired – the entity must evaluate whether the impairment is an OTTI.

An entity should recognize an OTTI on an impaired security when one of three conditions exists:

  1. The entity intends to sell the security
  2. It is more likely than not the entity will be required to sell the security prior to recovery of the amortized cost basis of the security
  3. The entity does not expect to recover the amortized cost basis of the security

If condition (1) or (2) exists, then the entity will reduce the amortized cost basis of the AFS debt security to its current fair value.  Any subsequent increases in the fair value of the AFS debt security would be recognized outside of earnings as a component of OCI until the gains are realized via cash collection or sale.

If neither condition (1) nor (2) exists, then the entity must evaluate whether it does not expect to recover the amortized cost basis of the security.  The entity may perform a qualitative analysis, considering factors such as the magnitude of the impairment, the duration of the impairment, factors relevant to the issuer of the security, factors relevant to the industry in which the issuer of the security operates, and any other relevant information.  Alternatively, an entity may perform a quantitative analysis by comparing the net present value (NPV) of expected cash flows of the AFS debt security to its amortized cost basis, as described below.

If the entity does not expect to recover the amortized cost basis of the security, an OTTI exists and the security should be written down to its fair value.  The entity must then separate the total impairment (the amount by which the AFS debt security’s amortized cost exceeds its fair value) between the amount of impairment related to (a) credit losses and (b) all other factors.  To make this distinction, the entity compares the NPV of the expected future cash flows on the debt security, discounted at the security’s effective interest rate (or “EIR”), to the amortized cost basis of the security.   The amount by which the amortized cost of the AFS debt security exceeds its NPV is recognized in earnings as a credit loss, while any remaining impairment is recognized outside of earnings as a component of OCI.

AFS Framework upon adopting ASU 2016-13

ASU 2016-13 largely keeps the OTTI framework from Subtopic 320-10 intact.  If either (1) an entity intends to sell, or (2) it is more likely than not that it will be required to sell an AFS debt security whose amortized cost exceeds its fair value, the entity shall write that AFS debt security’s amortized cost basis down to its fair value through earnings.  For AFS debt securities that are impaired, but for which neither (1) the entity intends to sell, nor (2) it is more likely than not that it will be required to sell an AFS debt security whose amortized cost exceeds its fair value, the entity will still need to assess whether it expects to recover the amortized cost basis of the impaired AFS debt security either via a qualitative analysis or via the same quantitative framework in Subtopic 320-10 today (as described above).

However, ASU 2016-13 makes a few important changes.  The most significant changes include:

  • Entities may no longer consider the duration of an impairment when qualitatively assessing whether the entity does not expect to recover the amortized cost basis of an impaired AFS debt security.
  • If an entity recognizes a credit loss on an AFS debt security, the entity will establish an allowance for credit loss (or “ACL”) rather than perform a direct write-down of the amortized cost basis of the AFS debt security. Accordingly, subsequent reductions in the estimated ACL will be recognized in earnings as they occur.
  • The amount of credit losses to be recognized is limited by a “fair value floor” – that is, total credit losses cannot exceed the total amount by which the amortized cost of the AFS debt security exceeds its fair value.

The following flow chart illustrates the how an entity would evaluate an AFS debt security for impairment upon adoption of ASU 2016-13:

Example

blog-chart

Background

  • Entity A has an investment in an AFS debt security issued by Company X with an amortized cost of $100
  • At 12/31/X1, the fair value of the AFS debt security is $90
  • The of the AFS debt security is 10% (as determined in accordance with ASC 310-20)

Entity A does not intend to sell the AFS debt security, nor is it more likely than not that Entity A will be required to sell the AFS debt security prior to recovery of the amortized cost basis.  Entity A elects to perform a qualitative analysis to determine whether the AFS debt security has experienced a credit loss.  In performing that qualitative assessment, Entity A consider the following:

  • Extent of impairment: 10%
  • Adverse conditions: Company X is in an industry that is in decline
  • Company X’s credit rating was recently downgraded

Accordingly, Entity A determines that a credit loss has occurred.  Next, Entity A makes its best estimate of expected future cash flows, and discounts those cash flows to their NPV at the AFS debt security’s EIR of 10% as follows:

In this case, the NPV is $85, which would indicate a $15 ACL.  However, the fair value of the AFS debt security is $90, so the ACL is limited to $10 due to the “fair value floor”.  Accordingly, Entity A would recognize a credit loss expense of $10 and create an ACL, also for $10.

In subsequent periods, Entity A would continue to determine the NPV of future expected cash flows and adjust the ACL up or down as those changes occur, subject to the fair value floor.[/vc_column_text][/vc_column][/vc_row][vc_row][vc_column][vc_empty_space][startapp_block_title animation=”” title=”About the Author”][/vc_column][/vc_row][vc_row][vc_column width=”1/6″][vc_single_image image=”2439″][/vc_column][vc_column width=”5/6″][vc_column_text]Graham Dyer, CPA Grant Thornton

Graham is a partner in Grant Thornton, LLP’s national office where he provides technical accounting guidance to clients across the globe.  Graham has a particular focus on financial institutions, including matters such as the ALLL, consolidations, Purchased Credit Impaired loan income recognition, complex financial instruments, business combinations, and SOX/FDICIA matters. ​

Graham also serves on a number of industry technical committees, including the IASB’s IFRS 9 Impairment Transition Group and the FASB’s CECL Transition Resource Group.  Graham was previously a professional accounting fellow at the OCC.


Fannie Mae and Freddie Mac Launch New Uniform Mortgage-Backed Security (UMBS)

Today, Fannie Mae and Freddie Mac begin issuing the long-awaited Uniform Mortgage-Backed Security (UMBS). The Federal Housing Finance Administration (FHFA) conceived of this new standard in its 2012 “A Strategic Plan for Enterprise Conservatorships,” which marked the start of the Single Security Initiative (the history of which is laid out in the graphic below). 

RiskSpan produces FHFA’s quarterly performance reports, most recently published Wednesday, May 29, which will support the agency’s oversight of the UMBS. The FHFA uses this report to monitor prepayment performance of passthroughs issued by Fannie and Freddie. These reports provide market participants with additional transparency on prepayment behavior alignment. They also allow the FHFA to monitor and address differences in conditional prepayments rates (CPR) between the two issuers and to align programs, policies, and practices that affect the cash flows of “To-Be-Announced” (TBA)-eligible Mortgage-Backed Securities (MBS). 

 The importance of RiskSpan’s contributions to the FHFA’s efforts are highlighted in Bloomberg’s May 30 article, “A $4 Trillion Plan Could Make or Break Dreams of U.S. Homebuyers”.


RiskSpan Credit Risk Transfer Solution

RiskSpan Managing Director, Janet Jozwik, explains how the RS Edge Platform serves as an end-to-end Credit Risk Transfer (CRT) solution designed to help investors in each stage of CRT deal analysis. The RS Edge Platform hosts historical GSE data (STACR/CAS/CIRT/ACIS) and gives users the ability to conduct historical and surveillance analysis as well as predictive and scenario analysis. Additionally, RiskSpan gives users full access to our proprietary agency-specific prepayment and credit models and is integrated with Intex for deal cash flow analysis.


Low MI No Problem: Analyzing the Historical Performance of Home Affordable Loans

Introduction

In our last CRT Deal Monitor post, we touched on a trend we have noticed- that the number of loans being originated with less-than-standard MI coverage has been increasing. This is a trend we will be covering in a series of blog posts. The following analysis provides a historical view of the performance of loans with less than standard MI coverage, like those being originated through the Fannie Mae HomeReady and Freddie Mac HomePossible programs. Fannie Mae CAS Deals contain a steadily growing percent of UPB in the HomeReady program. While Freddie Mac does not currently include a HomePossible indicator we suspect the same trend is occurring. In the coming months Freddie Mac will add this disclosure enhancement and we will investigate.

Historical data indicates that these HomeReady loans perform just as well, if not better, than similar loans not in an affordability program (see appendix for the cohort definitions). However, this trend appears to be shifting as newer vintages with standard MI have experienced less (albeit slightly) losses than their HomeReady counterparts, though there is significantly less performance history available. The table below shows the cumulative default rate for each vintage segmented by LTV cutoffs for the HomeReady Program.

Analysis

The plots below present a profile of Fannie Mae HomeReady and Standard MI cohorts via the distributions of UPB, LTV, FICO, and DTI dating back to 1999. The cohorts are similar, though the Standard MI cohort does present a slightly better credit profile. The Standard MI cohort contains more loans with <= 95% LTV, slightly higher FICOs, slightly lower DTIs, and higher average loan sizes.

All plots in this post are interactive:

  • Click and drag in any of the plots to zoom on a region.
  • Isolate groups by double clicking on the legend entries, and single click to add groups back in.

Cohort Characteristics Plots: To compare performance through time each cohort has been grouped by Vintage. The plot below shows the cumulative default rate based on months from origination for each Vintage MI cohort. Based on the data, the older HomeReady population has experienced a lower overall default rate vs. the same vintage with Standard MI. This effect is exaggerated for vintages originated immediately preceding the crisis and is observed consistently through 2011.

Unsurprisingly, since the Low MI cohorts experienced a lower overall default rate, they also experienced a lower cumulative net loss which is displayed for each vintage on hover. Select a single vintage from the dropdown menu or isolate vintage(s) by clicking the lines or legend.

Cumulative Default Rate Plot: Since the HomeReady population is characterized by having less than standard MI, we should expect this population to have a higher loss severity. This relationship is seen in the data and is most prominent from the 2005 vintage onward. With the exception of the 2011 vintage, the gap between severity for Low and Standard MI has grown stronger through time.

Cumulative Severity Plot: In the next installment of this series we will cover specific loss characteristics for the HomeReady and Standard MI populations, and discuss the impact of Borrower Area Median Income, which is an eligibility requirement for the HomeReady population.

Appendix:

Cohort Selection Criteria:

For this analysis, the historical performance of two cohorts ‘Low MI’ and ‘Standard MI’ were pulled from RiskSpan’s Edge Platform from the Fannie Mae Loan Performance Dataset. The cohorts contain approximately 800,000 and 2,1M loans respectively. The cohorts were established based on the current MI coverage requirements set by Fannie Mae, and were limited to loans with LTV > 90.1%. The matrix below shows MI coverage requirements for the HomeReady (Low MI) cohort and Standard MI cohort.

Cohort 1 – Low MI Coverage:

Cohort 2 – Standard MI Coverage:


Automate Your Data Normalization and Validation Processes

Robotic Process Automation (RPA) is the solution for automating mundane, business-rule based processes so that organizations high value business users can be deployed to more valuable work. 

McKinsey defines RPA as “software that performs redundant tasks on a timed basis and ensures that they are completed quickly, efficiently, and without error.” RPA has enormous savings potential. In RiskSpan’s experience, RPA reduces staff time spent on the target-state process by an average of 95 percent. On recent projects, RiskSpan RPA clients on average saved more than 500 staff hours per year through simple automation. That calculation does not include the potential additional savings gained from the improved accuracy of source data and downstream data-driven processes, which greatly reduces the need for rework. 

The tedious, error-ridden, and time-consuming process of data normalization is familiar to almost all organizations. Complex data systems and downstream analytics are ubiquitous in today’s workplace. Staff that are tasked with data onboarding must verify that source data is complete and mappable to the target system. For example, they might ensure that original balance is expressed as dollar currency figures or that interest rates are expressed as percentages with three decimal places. 

Effective data visualizations sometimes require additional steps, such as adding calculated columns or resorting data according to custom criteria. Staff must match the data formatting requirements with the requirements of the analytics engine and verify that the normalization allows the engine to interact with the dataset. When completed manually, all of these steps are susceptible to human error or oversight. This often results in a need for rework downstream and even more staff hours. 

Recently, a client with a proprietary datastore approached RiskSpan with the challenge of normalizing and integrating irregular datasets to comply with their data engine. The non-standard original format and the size of the data made normalization difficult and time consuming. 

After ensuring that the normalization process was optimized for automation, RiskSpan set to work automating data normalization and validation. Expert data consultants automated the process of restructuring data in the required format so that it could be easily ingested by the proprietary engine.  

Our consultants built an automated process that normalized and merged disparate datasets, compared internal and external datasets, and added calculated columns to the data. The processed dataset was more than 100 million loans, and more than 4 billion recordsTo optimize for speed, our team programmed a highly resilient validation process that included automated validation checks, error logging (for client staff review) and data correction routines for post-processing and post-validation. 

This custom solution reduced time spent onboarding data from one month of staff work down to two days of staff work. The end result is a fullyfunctional, normalized dataset that can be trusted for use with downstream applications. 

RiskSpan’s experience automating routine business processes reduced redundancies, eliminated errors, and saved staff time. This solution reduced resources wasted on rework and its associated operational risk and key-person dependencies. Routine tasks were automated with customized validations. This customization effectively eliminated the need for staff intervention until certain error thresholds were breached. The client determined and set these thresholds during the design process. 

RiskSpan data and analytics consultants are experienced in helping clients develop robotic process automation solutions for normalizing and aggregating data, creating routine, reliable data outputsexecuting business rules, and automating quality control testing. Automating these processes addresses a wide range of business challenges and is particularly useful in routine reporting and analysis. 

Talk to RiskSpan today about how custom solutions in robotic process automation can save time and money in your organization. 


RiskSpan VQI: Current Underwriting Standards – Quarter 1 2019

q1 vqi

The RiskSpan Vintage Quality Index (“VQI”) rose to its highest point (105.16), since September of 2008, in January of 2019, before dropping down to just below 100 in March of 2019. The spike in January was due to a slight increase in all risk factors with cash-out refinances going up by 1.4%. The drop in the VQI from January to March was due to a 5.7% decrease in LTV’s over 80, a 4.2% decrease in DTI’s over 45, and a 1.1% decrease in FICO’s under 660. However, cash-out refinances continued its upward trend and increased by 2.1% in the same time period.

RiskSpan introduced the VQI in 2015 as a way of quantifying the underwriting environment of a particular vintage of mortgage originations. The idea is to provide credit modelers a way of controlling for a particular vintage’s underwriting standards, which tend to shift over time. The VQI is a function of the average number of risk layers associated with a loan originated during a given month. It is computed using the loan-level historical data released by the GSEs in support of their Credit Risk Transfer initiatives (CRT data) for months prior to December 2005, and using loan level disclosure data supporting MBS issuances through today. The value is then normalized such that January 1, 2003 has an index value of 100. The peak of the index, a value of 139 in December 2007, indicates that loans issued in that month had an average risk layer factor 39% greater (i.e., loans issued that month were 39% riskier) than loan originated during 2003. In other words, lower VQI values indicate tighter underwriting standards (and vice-versa).

Build-Up of VQI

The following chart illustrates how each of the following risk layers contributes to the overall VQI:

  • Loans with low credit scores (FICO scores below 660)
  • Loans with high loan-to-value ratios (over 80 percent)
  • Loans with subordinate liens
  • Loans with only one borrower
  • Cash-out refinance loans
  • Loans secured by multi-unit properties
  • Loans secured by investment properties
  • Loans with high debt-to-income ratios (over 45%)
  • Loans underwritten based on reduced documentation
  • Adjustable rate loans

Analytical and Data Assumptions

Population assumptions:

  • Issuance Data for Fannie Mae and Freddie Mac.
  • Loans originated more than three months prior to issuance are excluded because the index is meant to reflect current market conditions.
  • Loans likely to have been originated through the HARP program, as identified by LTV, MI coverage percentage, and loan purpose are also excluded. These loans do not represent credit availability in the market, as they likely would not have been originated today if not for the existence of HARP.

Data Assumptions:

  • Freddie Mac data goes back to December 2005. Fannie Mae data only goes back to December 2014.
  • Certain Freddie Mac data fields were missing prior to June 2008.

GSE historical loan performance data release in support of GSE Risk Transfer activities was used to help back-fill data where it was missing.

Get a Demo

Note: The analysis in this blog post was developed using RiskSpan’s Edge Platform. The RiskSpan Edge Platform is a module-based data management, modeling, and predictive analytics software platform for loans and fixed-income securities. Click here to learn more.


Get Started