Foundations · HMDA
How to read HMDA mortgage data without misreading it
Public mortgage records can show what lenders reported in a past year. They cannot tell you whether one person will be approved—or why two groups or places differ without further analysis.
The short answer
Read every HMDA percentage as a description of a defined set of lender-reported records in one activity year. Check the denominator, geography, loan purpose, and outcome before making a comparison. Never read a historical share as a personal approval probability.
01 · The dataset
What HMDA is
The Home Mortgage Disclosure Act requires many financial institutions to maintain, report, and publicly disclose information about mortgage lending. The public data describes applications, originations, purchased loans, and other reported outcomes for a calendar activity year.
HMDA is unusually detailed, but it is not a live rate feed, a credit report, or a complete record of every mortgage-related transaction in the United States. Public files are also modified to protect applicant privacy.
02 · The essentials
Six numbers to understand first
- Activity year
- The calendar year covered by the lender’s submission—not the date you opened the page.
- Applications
- Records for outcomes 1, 2, 3, 4, 5, 7, and 8. MortgagesVine excludes purchased loans because they were not applications submitted to that reporting institution.
- Originations
- Applications that resulted in a loan being made. This is one outcome, not a measure of all approvals.
- Reporting lenders
- The institutions represented in the selected place and year—not necessarily every lender operating there.
- Outcome share
- An outcome count divided by the application denominator used on that page.
- Loan purpose
- Why the credit was requested, such as home purchase, refinancing, cash-out refinancing, or home improvement.
03 · Reading a result
Outcome labels are not interchangeable
A loan can fail to originate for several different reasons. Keep each reported action separate rather than treating every non-origination as a denial.
04 · Comparison
Compare like with like
A difference becomes interpretable only after the comparison is aligned. Before comparing two lenders, counties, metros, or years, check all six conditions.
- Same activity year
- Same geography level
- Same loan purpose and product
- Same outcome definition
- Enough observations to be stable
- Counts and denominators shown together
Even a carefully aligned comparison is descriptive. It may reveal a question worth investigating; it does not, by itself, establish the cause of a difference.
05 · Boundaries
What HMDA cannot tell you
The applicants, products, prices, underwriting policies, and economic conditions in a past dataset may differ from your situation today.
- It does not provide a personalized approval forecast.
- It does not show every factor a lender considered.
- It does not prove discrimination or explain causation on its own.
- It does not tell you which lender offers the best current price.
- It does not replace a Loan Estimate, legal advice, or housing counseling.
06 · Reported reasons
Denial reasons need their own denominator
MortgagesVine county pages divide each reported reason count by all application and preapproval denial decisions. A lender may report more than one reason for one decision, so the shares can add to more than 100%. Missing reasons are not inferred, and small counts are suppressed.
Treat labels such as debt-to-income ratio, credit history, or collateral as categories reported in HMDA—not as a complete explanation of an individual underwriting decision.
07 · Go to the record