Most people would agree that 77 million – approximate the number of loan applications submitted pursuant to the Home Mortgage Disclosure Act of 1975 as amended (“HMDA”) for the period 2010 through 2014 – is a big number.
The number of data fields currently provided for each loan application – 43 – isn’t a particularly big number. The expected number of additional HMDA data fields likely to be required by the Consumer Financial Protection Bureau – approximately 20 – isn’t particularly large either.
The possible values for the current 43 fields – including 74,134 census tracts, more than 5,000 discrete income values and more than 10,000 discrete loan amounts – are large numbers but not difficult to imagine.
Put all those numbers together – 77 million applications, 43 fields, up to 74,134 discreet values in any one data field – and you have more than 200 trillion – that’s trillion – possible data permutations or metadata elements. To put this number in context, 200 trillion is more than 10 times larger than the national debt of the United States of America.
200+ trillion metadata elements. The enormity of this number and the limitations of analytical tools such as Excel and even basic query and reporting functionality of relational database products (i.e., SQL) makes the meaningful use of this data challenging. In light of these limitations, it’s not surprising that the value of HMDA data hasn’t really been unlocked. Until now.
Combining the right advanced analytical platform with HMDA data provides a unique opportunity for mortgage lenders to transition from “HMDA data” to “HMDA intelligent insight” through integrated advanced analytics and strategic benchmarking.
Integrated Advanced Analytics
Business intelligence is nothing more than the insight gained through the integration of advanced analytics and innovative visualization. Advanced analytics is simply one or more dynamic, multi-dimensional metrics that allow the user to explore proprietary and third-party financial and operational indicators based on need, experience, intuition, and real-time insights.
While business intelligence supports traditional enterprise benchmarking, HMDA data demonstrates how strategic benchmarking – comparing one or more key indicators for one institution to the same indicators for one or more other institutions – produces a level of insight that unifies revenue, risk, and compliance activities in a way that enhances institutional sustainability by highlighting whether an institution leads or lags the overall market.
Scoring and Benchmarking: Mortgage TrueView HMDA S|B
HMDA S|B provides mortgage lenders and other stakeholders with insight through the five publicly accessible proprietary scores and integrated dynamic benchmarking.
D Score. Measures lender decisiveness as expressed through the relationship between actionable applications (i.e., approvals and denials) and non-actioned applications (i.e., withdrawals and incompletes).
A Score. Indicates a lender’s relative rate of affirmative outcomes through the relationship between approved and denied applications.
B Score. Presents the ratio of an applicants’ reported income, where included, to applicants’ requested loan amount to provide an overall indication of ability- to repay.
E Score. Measures a lender’s level of engagement by comparing a lender’s percentage of non-white, non-male, applications to corresponding percentage of denied applications.
C Score. Compares the denial rates for twenty gender, race, and ethnicity groups relative to the denial rate for white, male, non-Latino applicants.
A detailed description of the HMDA S|B D Score is included within this document. The other HMDA S|B scores will be described in subsequent documents. HMDA S|B scores are clearly descriptive and therefore provide a critical overview of market results and trends. However, scores are recalculated as Respondent, year, loan type, MSA and other available filters are applied to the data. We believe this filtering ability not only provides critical insight at more actionable levels, but also tempers any criticism concerning the value of scoring such as vast dataset.