PRI Data Analytics Blog
Subscribe to get powerful insights
to help your business grow.
to help your business grow.
Polygon Research First With 2018 HMDA Data
August 19, 2019
Polygon Research has been working with HMDA data for several years, and has been selling subscriptions to HMDAVision, our Software-as-a-Service HMDA analysis app, since 2018. HMDAVision contains the most recent 5 years of HMDA data, so we are able to show not only the latest information, but trends over time as well.
The FFIEC, and more recently the CFPB, have traditionally released the prior year's HMDA data each spring in the form of a national loan level dataset consisting of a large (~2GB) LAR file along with supporting lender and geography information files. This year marks a departure from this pattern in that the LAR information has so far only been released as individual per-lender "Modified LAR" files without the supporting files. Per the CFPB: "The Modified LARs contains loan level information for 2018 on individual HMDA filers, modified to protect privacy." And the bar is even higher right now for organizations wanting to make sense of HMDA trends through 2018, as 2018 ushers in the use of a new unique identifier for lenders, with no crosswalk to the previous one.
The FFIEC, and more recently the CFPB, have traditionally released the prior year's HMDA data each spring in the form of a national loan level dataset consisting of a large (~2GB) LAR file along with supporting lender and geography information files. This year marks a departure from this pattern in that the LAR information has so far only been released as individual per-lender "Modified LAR" files without the supporting files. Per the CFPB: "The Modified LARs contains loan level information for 2018 on individual HMDA filers, modified to protect privacy." And the bar is even higher right now for organizations wanting to make sense of HMDA trends through 2018, as 2018 ushers in the use of a new unique identifier for lenders, with no crosswalk to the previous one.
Enter Polygon Research. We have a world-class technology platform thanks to our partnership with Qlik, and we are experts in the mortgage industry. We are also a bootstrapping, family-owned business focused on providing valuable services at fair prices. We decided to tackle the technology challenge of modeling the 5600+ modified LAR files along with the research challenge of producing our own crosswalk files because we believe the mortgage industry is better - more fair, more robust, and more efficient - when lenders, agencies, and all stakeholders are able to participate in the same dialogue. HMDAVision makes this possible by allowing all subscribers to drill down and cross filter by any combination of the 100+ Loan, Applicant, Housing Inventory, Lender, and Secondary Market dimensions and measures we include in the app.
HMDAVision: 2018 HMDA Transactions by Action Type. Note that Alternative Dimensions can be selected on the x axis, and Alternative Measures can be selected on the y axis.
Next for us is to continue to update HMDAVision with new data posted by the CFPB - both updates to the modified LARs, and then to the new national loan level dataset once released. This reflects a core additional advantage to our subscriptions: we provide continual updates to our HMDA and Census data as part of our flat subscription price so our customers always have the latest and greatest data.
Contact us with any questions about the 2018 HMDA data or to request a demo.
Contact us with any questions about the 2018 HMDA data or to request a demo.
HMDAVision Now Live with 2018 HMDA Data!
July 19, 2019
Since the CFPB first released its 2018 HMDA Data Home Mortgage Disclosure Act (HMDA) Modified Loan Application Registers (LARs) data earlier this year, Polygon Research has been working to acquire, profile, model, cross-reference, and visualize the data. We are really excited about the result – a major new release of our Software-as-a-Service app HMDAVision.
There are three highlights we'd like to share today, leaving more detail for future blog posts.
There are three highlights we'd like to share today, leaving more detail for future blog posts.
Scope
- HMDAVision addresses all 85 data fields of the available 2018 HMDA Modified LAR data for the more than 5,600 reporting lenders whose data meets our threshold for relevance and reportability.
- To this we add:
- over 2 dozen custom, calculated fields for enriched analysis
- microdata from the latest year's US Census American Community Survey (updated each Fall) for deep analysis of the population and housing inventory in the areas in which our customers are lending or working with lenders
Data Quality
The 2018 data looks great, with a couple exceptions (HMDAVision is updated once a month with any updates and corrections made available by the CFPB):
Example 1: 2018 Applicant Income has the lowest percentage of empty fields of the past 5 years, which augurs well for good data quality, but the incomes include negative values for the first time. This is consistent with definitions of gross income, but may be a surprise if you're not expecting it. The data also includes some quite high and quite low outliers. It's hard to know if outliers are accurate; HMDAVision helps by allowing you to filter out the applications which didn't reach a credit decision, and may therefore only be self-reported.
Example 2: There is a conflict in some cases between the FIPS code and the state abbreviation in the data – our decision in these cases has been to go with the more detailed information contained in the FIPS. HMDAVision is based on microdata – this allows granular filtering to see applications by lender, by MSA, County, or Census Tract. Or by any of the other dimensions in the data, in any combination. If you ever detect something you think is an anomaly in the data, let us know, and we'll investigate further.
Example 1: 2018 Applicant Income has the lowest percentage of empty fields of the past 5 years, which augurs well for good data quality, but the incomes include negative values for the first time. This is consistent with definitions of gross income, but may be a surprise if you're not expecting it. The data also includes some quite high and quite low outliers. It's hard to know if outliers are accurate; HMDAVision helps by allowing you to filter out the applications which didn't reach a credit decision, and may therefore only be self-reported.
Example 2: There is a conflict in some cases between the FIPS code and the state abbreviation in the data – our decision in these cases has been to go with the more detailed information contained in the FIPS. HMDAVision is based on microdata – this allows granular filtering to see applications by lender, by MSA, County, or Census Tract. Or by any of the other dimensions in the data, in any combination. If you ever detect something you think is an anomaly in the data, let us know, and we'll investigate further.
Value
The net new fields in the 2018 data include a ton of useful information:
Example 1: See our new Reg Z Pricing sheet (each of the 13 sheets in our app has a theme to which we dedicate several individual charts, tables, and maps). Understand lender performance relative to peer groups which you define and control.
Example 2: We model 37 fields across all 5 years in HMDAVision, with 60 fields unique to 2018. We're excited about the significant value-add provided by these new, 2018-specific fields!
Example 1: See our new Reg Z Pricing sheet (each of the 13 sheets in our app has a theme to which we dedicate several individual charts, tables, and maps). Understand lender performance relative to peer groups which you define and control.
Example 2: We model 37 fields across all 5 years in HMDAVision, with 60 fields unique to 2018. We're excited about the significant value-add provided by these new, 2018-specific fields!
Driving Insights from the American Community Survey
June 11, 2019
Panelists discussing the implications of the study's findings in the Brookings Institute's Faulk Auditorium.
On May 20th, Polygon Research, Inc. attended the unveiling of a fascinating research paper on the state of housing growth and affordability in the capital region at the Brookings Institution. This research paper was the product of a joint effort among the Brookings Institute, Fannie Mae, Georgetown University, and George Washington University. These organizations carried out a focused analysis of U.S. housing trends over time using Census Bureau American Community Survey (ACS) data sets from 2000 to 2017. The study assessed differences in housing developments according to the urban, suburban, and exurban segments and their respective counties in the D.C. metropolitan area. It also explored the different patterns between renter vs. homeowner status as well as housing value vs. income.
A key finding from the analysis pointed to the disproportionate growth rate of housing in exurban counties at 45% growth compared to a combined growth rate of 15% in suburban and urban counties. Moreover, the data revealed a relative increase in the percentage makeup of homeowner-occupied housing from urban to suburban to exurban areas, respectively rising from 40% to 65% to 75%. This highlights the understanding that D.C.'s urban areas consist of a greater concentration of renters at 60% versus suburban and exurban regions at around 30%. PRI was also interested in the report's finding that the median housing value was more than four times the median income across the entire Washington MSA, yet the ratio of housing value to income fluctuated greatly according to segment, ranging from a 7:1 ratio in urban areas to a 3:1 ratio in exurban areas.
The paper visualized these key findings and data trends through charts and maps, comprehensively displaying the concentration of an outcome by county. To check out these insights and read more about housing and affordability in the D.C. metropolitan area, click here.
A key finding from the analysis pointed to the disproportionate growth rate of housing in exurban counties at 45% growth compared to a combined growth rate of 15% in suburban and urban counties. Moreover, the data revealed a relative increase in the percentage makeup of homeowner-occupied housing from urban to suburban to exurban areas, respectively rising from 40% to 65% to 75%. This highlights the understanding that D.C.'s urban areas consist of a greater concentration of renters at 60% versus suburban and exurban regions at around 30%. PRI was also interested in the report's finding that the median housing value was more than four times the median income across the entire Washington MSA, yet the ratio of housing value to income fluctuated greatly according to segment, ranging from a 7:1 ratio in urban areas to a 3:1 ratio in exurban areas.
The paper visualized these key findings and data trends through charts and maps, comprehensively displaying the concentration of an outcome by county. To check out these insights and read more about housing and affordability in the D.C. metropolitan area, click here.
PRI was excited to see the use of ACS microdata to shed light on housing trends and gaps in affordability within the capital region. Polygon Research's interactive apps, available via subscription, also incorporate ACS microdata that allows researchers, consumer groups, and the general public to find and communicate housing and population insights and trends for any region in the United States.
Take a look at how PRI harnesses the power of 5-year ACS data analysis and visualization in its soon-to-be-released CensusVision app, which complements and expands upon the paper's findings:
Take a look at how PRI harnesses the power of 5-year ACS data analysis and visualization in its soon-to-be-released CensusVision app, which complements and expands upon the paper's findings:
There is a significantly greater disparity in median incomes between homeowners and renters within D.C. (left) as opposed to the aggregate metropolitan area (right).
Qlik Federal Summit 2019
June 7, 2019
HMDAVision poster on display at the Qlik Federal Summit June 6, 2019
The 2019 Qlik Federal Analytics Summit in Washington, DC showcased the evolution Qlik has been going through over the past two years, which has created a really compelling value proposition:
In addition to Qlik's own content, this event is always a great chance to see the great things the federal government and non-profits (e.g. Office of the Undersecretary of Defense; GSA; Mercy Ships) are achieving. They are the embodiment of Qlik CEO Mike Capone's entreaty that if you want to compare Qlik to its competitors, choose a hard challenge you're facing and real outcomes you are trying to achieve. To everyone we met yesterday: keep up the great work!
- Data Literacy mission―Qlik's campaign to democratize data is in its second year and going strong. It is at the heart of a virtuous cycle to both improve society (literacy is good!) and advance Qlik's brand.
- Adept moves regarding scalability and portability―Qlik Sense on Kubernetes; true portability across public or private cloud, on-premise, and SaaS (Qlik or managed service provider).
- Impressive enhancements in the analytics engine that powers Qlik Sense―rich APIs; Qlik Core; Associative Big Data Index.
- Solid ecosystem―Qlik recently purchased Qlik Data Catalyst (Podium Data) for data management and data discovery, and Attunity for data movement and data integration. These new products – now with deeper hooks into Qlik Sense, while retaining their heterogeneous integration capabilities – paint a compelling end-to-end picture for data architects, while squarely communicating to the C-suite that their investment in Qlik won't paint them into a corner.
In addition to Qlik's own content, this event is always a great chance to see the great things the federal government and non-profits (e.g. Office of the Undersecretary of Defense; GSA; Mercy Ships) are achieving. They are the embodiment of Qlik CEO Mike Capone's entreaty that if you want to compare Qlik to its competitors, choose a hard challenge you're facing and real outcomes you are trying to achieve. To everyone we met yesterday: keep up the great work!
2019 BI Trends
February 4, 2019
We are proud to present to you this exciting webinar about the state of analytics and trends in 2019, delivered by Dan Sommer, Qlik's Market Intelligence Lead. You will learn about:
- Emerging approaches to AI, data literacy, and embedded analytics
- Rising trends in infrastructure, service delivery, and data management
- What postmodern analytics will look like ― and why we need it now
Right Tool to Handle Expanded Public HMDA Data
January 15, 2019
For years, spreadsheets have been the main tool by which HMDA data is analyzed and consumed, but the breadth and depth of analysis they have offered in the past is no longer keeping up with the volume and velocity of the expanded public HMDA data set in 2019.
At the end of Q1, the public HMDA data set will have data fields describing mortgage loan applications in greater detail with over 25 data points per loan – most of which will be disclosed without modification.
The data on the borrower will be enriched with additional information for Asian and Hispanic applicants/borrowers. The data on property used to secure the loan will also be expanded to include the manufactured home market in addition to single-family and multifamily homes. For the first time, data users and the public will be able to see info on AUS, credit score models, modified credit scores, and more.
The data on the borrower will be enriched with additional information for Asian and Hispanic applicants/borrowers. The data on property used to secure the loan will also be expanded to include the manufactured home market in addition to single-family and multifamily homes. For the first time, data users and the public will be able to see info on AUS, credit score models, modified credit scores, and more.
Will the lenders and other interested parties be able to ingest and make sense of this detailed data with traditional spreadsheets?
Spreadsheets are limited in their capacity to open single, large files, and even more so in their ability to blend multiple data sources. These limitations prevent institutions from taking full advantage of the market insights available in the HMDA data at best, and at worst, they put institutions at risk with inaccurate conclusions about fair lending.
Modern BI and data analytics tools are needed to ingest large data sets and build an accurate picture of where the industry is and where it is going. An effective data analytics solution streamlines complex analytic modeling and, most importantly, delivers answers so lenders and other stakeholders can adjust their strategies to capture market opportunities. Polygon Research's HMDAVision – deployed as-a-service in the cloud – leapfrogs other solutions by providing fully modeled, ready for analysis, fully interactive insights into the HMDA data from the last 5 years. HMDAVision delivers immediate insights into the market: trends, lending strategies, and compliance, and can be leveraged even further through integrations with other platforms and data.
Marketing Strategy Accelerator:
Benchmark any lender's products and underwriting practices against its peers in any market or product category
Sales Strategy Accelerator:
Uncover the whole story within the expanded HMDA data about your strengths in your target markets
Stakeholder Communications Accelerator:
Proactively share insights with your organization, customers, regulators, and other stakeholders
Hiring Accelerator:
Construct a smart, efficient talent acquisition strategy by understanding the distribution of MLO talent by geography and by business
Future Accelerator:
Run a variety of "what-if" market research models to explore trends in any segment of the market and to chart your course into the future
Responding to the HMDA Data Challenge - Announcing HMDAVision
October 22, 2018
HMDAVision contributes to the mortgage banking industry in two important ways:
1) By providing a powerful common analytics platform that can be accessed at the same time by all stakeholders in the industry who will see the same information, HMDAVision facilitates a constructive, fact-based conversation about lending practices and trends.
2) Delivered as-a-service, HMDAVision dramatically lowers costs of analysis of this data and becomes a low-cost and efficient alternative for market analysis.
1) By providing a powerful common analytics platform that can be accessed at the same time by all stakeholders in the industry who will see the same information, HMDAVision facilitates a constructive, fact-based conversation about lending practices and trends.
2) Delivered as-a-service, HMDAVision dramatically lowers costs of analysis of this data and becomes a low-cost and efficient alternative for market analysis.
Polygon Research is proud to announce the launch of its first cloud-based application for the residential mortgage industry: HMDAVision. HMDAVision aims to revolutionize the way HMDA data is analyzed and consumed. We have done the hard work of modeling the latest five years of HMDA data and of presenting it in interactive charts which objectively present KPIs and trends.
Insights available:
Market Size & Growth – Understand key dynamics by geography, product, borrower, economic area, property type and lien status
Housing Market Overview – Instantaneous insight into the nature of housing stock, and the size of the housing market
Population Overview – Gain quick understanding of the demographic makeup of the market and the market originations by borrower segment
Market Health and Distribution – Get the pulse of the health of the market
Competitive Landscape – Define your market and get immediate answers about the market players
Lender Benchmarks – Understand peers in market: analyze their production volumes, and glean their marketing, sales, and underwriting strategies by a variety of measures
Denials – Understand denial trends and see primary, secondary, and tertiary denial reasons. Drill down and cross filter to analyze your and your competitors' patterns. Cross filter by all major HMDA dimensions
Closing Rate – Understand the trends in closing rate by a variety of measures and dimensions, including geography, type of lender, borrower segment, product segment, and many others
Insights available:
Market Size & Growth – Understand key dynamics by geography, product, borrower, economic area, property type and lien status
Housing Market Overview – Instantaneous insight into the nature of housing stock, and the size of the housing market
Population Overview – Gain quick understanding of the demographic makeup of the market and the market originations by borrower segment
Market Health and Distribution – Get the pulse of the health of the market
Competitive Landscape – Define your market and get immediate answers about the market players
Lender Benchmarks – Understand peers in market: analyze their production volumes, and glean their marketing, sales, and underwriting strategies by a variety of measures
Denials – Understand denial trends and see primary, secondary, and tertiary denial reasons. Drill down and cross filter to analyze your and your competitors' patterns. Cross filter by all major HMDA dimensions
Closing Rate – Understand the trends in closing rate by a variety of measures and dimensions, including geography, type of lender, borrower segment, product segment, and many others
How Can Sales Teams Benefit from HMDAVision?
Leverage analytical maps to reveal gaps in originations at geographic market, and drill into these gaps by property type, product, type, borrower ethnicity, and other dimensions.
Quantify the market opportunity to help lenders or individual bankers/brokers grow efficiently.
Use peer benchmarking data to position and communicate business strategy when recruiting new loan officers or to inform loan officer compensation plans.
Identify sales expansion opportunities – target branches and MLOs
How Can Compliance Teams Benefit from HMDAVision?
Monitor areas of concern with HMDAVision Risk Indicators:
Underwriting: Monitor Denial Rates and other Measures
Pricing: Monitor the disparities in the number of higher-priced on a prohibited basis
Redlining: Consider the lack of applications from lending in minority areas and disproportionately high denial rates for applicants located in minority areas
Steering: Monitor the gaps between percentages of prohibited basis groups in each of the loan alternative products
Marketing: Lower levels of applications from prohibited basis groups compared to their representation in the total population of the bank's market area
Underwriting: Monitor Denial Rates and other Measures
Pricing: Monitor the disparities in the number of higher-priced on a prohibited basis
Redlining: Consider the lack of applications from lending in minority areas and disproportionately high denial rates for applicants located in minority areas
Steering: Monitor the gaps between percentages of prohibited basis groups in each of the loan alternative products
Marketing: Lower levels of applications from prohibited basis groups compared to their representation in the total population of the bank's market area
Roadmap
In 2019, the industry will produce an enriched HMDA data set. The Bureau of Consumer Financial Protection (BCFP) is still in the process of deciding which fields will be released to the public. Even if the BCFP decides to release only few of the additional data points, the conversation among stakeholders will be enriched. Polygon Research plans to establish and continually update HMDAVision so it can serve as a reliable tool for all US residential mortgage industry stakeholders.
FinTech Lending Trends
March 19, 2018
HMDA data, published each year for the preceding year, shows interesting trends over time
Technological innovation has improved the efficiency of financial intermediation in the U.S. mortgage market. And whether we want to admit it or not, technology has continued to reshape the mortgage market place. A quick glance at 2016 transaction-level HMDA data tells us that two of the top ten lenders in the U.S. are online lenders (Quicken Loans and LoanDepot). The side-by-side geo-spatial comparison of their combined market share shows significantly deeper market penetration in 2016 as compared to 2012.
How did FinTech Lenders Achieve This?
It is difficult to answer this question with certainty, but there are some undeniable market developments that have contributed to the rise of such lenders. For example, one explanation is the confluence of several trends:
In 2016, half of the top 10 lenders in the U.S. were non-bank lenders with pure online lending business models or with serious online technology offerings, two were smaller banks with significant correspondent business, and only three were big national banks (see the bar chart for details). In 2017, Quicken Loans announced that it surpassed Wells Fargo in mortgage originations.
- a regulatory environment that drives banks to be risk-averse and retreat from lending to riskier borrowers
- an unprecedented rise in computing power and processing speed
- the availability of mobile applications to a wide range of borrowers
- millennials' entry into the mortgage market
- a deliberate strategy on the part of some non-bank lenders to drive volume through technology innovation that produces efficiencies and cost savings while increasing the lender footprint
In 2016, half of the top 10 lenders in the U.S. were non-bank lenders with pure online lending business models or with serious online technology offerings, two were smaller banks with significant correspondent business, and only three were big national banks (see the bar chart for details). In 2017, Quicken Loans announced that it surpassed Wells Fargo in mortgage originations.
What are Some of the Typical Characteristics of FinTech Lenders?
In February, a FRBNY staff report showed that:
Clearly, for mortgage lenders who are looking to grow, applying innovative technology and agile processes in every step of the loan cycle is not an option but a necessity.
For insights and business intelligence solutions for the mortgage marketplace, contact Polygon Research, Inc.
- FinTech lenders process mortgage applications about 20 percent faster than other lenders, even when controlling for detailed loan, borrower, and geographic variables.
- FinTech lenders are able to adjust supply more elastically than other lenders in response to mortgage demand shocks.
- FinTech lenders are able to alleviate capacity constraints associated with traditional mortgage lending.
Clearly, for mortgage lenders who are looking to grow, applying innovative technology and agile processes in every step of the loan cycle is not an option but a necessity.
For insights and business intelligence solutions for the mortgage marketplace, contact Polygon Research, Inc.
Federal HMDA Reporting Agency Landscape
March 5, 2018
HMDA data, published each Fall for the preceding year, shows interesting trends over time
The HMDA reporting agency landscape has changed over the last five years. As non-bank lenders filled the void in the retail mortgage originations market left by depository institutions, HUD swapped places with CFPB as the largest of the agencies by loan volume, and NCUA surpassed the OCC. With respect to loan product mix, the biggest game changer for HUD lenders was a significant increase in VA loan volume.
HMDA for Market Sizing
February, 25 2018
Are you using HMDA data for your strategic planning?
Home Mortgage Disclosure Act (HMDA) Data, gathered and published annually by FFIEC, has been used by regulators and interest groups to scrutinize mortgage lenders for compliance gaps. As a result, many lenders view HMDA Data as a threat. However, lenders would be wise to take a second look. HMDA Data is rich with market insights that can guide decision-making. See our Market Sizing example below to understand some of the questions HMDA can help to answer.
Polygon Research's President Recognized Among 40 New Certified Mortgage Bankers
October 23, 2017
