Use your widget sidebars in the admin Design tab to change this little blurb here. Add the text widget to the Blurb Sidebar!
Home Commentary Metropolitan Spotlights Dashboard Data Maps Blog Subscribe to MetroTrends Blog - RSS RSS icon

What HMDA data say about the state of the mortgage market

Author: Ellen Seidman

| Posted: October 1st, 2013

 
 

housesIcon

Autumn brings crisper nights, apple cider, and the annual release of the prior year’s Home Mortgage Disclosure Act (HMDA) data.

HMDA requires the vast majority of home mortgage lenders to disclose critical information about each loan application, such as whether the loan was approved or denied, and information about the loans and applicants—including the race, gender, and income of the borrower and the mortgage’s location.

While regulatory use of HMDA focuses heavily on each lender’s service to its community (for banks, performance under the Community Reinvestment Act or CRA; for all lenders, concerns about discrimination), HMDA’s vast dataset—2012’s included 9.8 million loans reported by 7,400 lenders— provides important insight into the state of the mortgage market.

The Federal Reserve Board of Governors’ annual HMDA analysis, authored by Neil Bhutta and Glenn Canner. “Mortgage Market Conditions and Borrower Outcomes: Evidence from the 2012 HMDA Data and Matched HMDA-Credit Record Data” is especially notable this year because it includes analysis of the performance through 2012 of 300,000 loans made in 2006 (at the height of the bubble and of sloppy underwriting) and in 2010 (when mortgage credit was extremely tight).

This ability to follow a mortgage is an important innovation, attributable to an ongoing project called the National Mortgage Database (NMDB), a joint effort of the Federal Housing Finance Administration and Freddie Mac.  (The NMDB was featured at the June data lunch sponsored by the Urban Institute’s new Housing Finance Policy Center.)

Here’s what we learn from the basic 2012 HMDA analysis:

  • New mortgage loans increased 38 percent from 2011 to 2012, but that was largely driven by a 54 percent increase in refinancing loans, compared to only a 13 percent increase in home purchase loans.
  • Government financing—Federal Housing Agency (FHA), Veterans Administration (VA), and Rural Housing Service (RHS)—accounted for 45 percent of the home purchase loans for owner-occupants, with VA and RHS lending particularly strong. The median income of borrowers who got home purchase loans from the FHA was about 40 percent lower than the median for conventional (non-government) home purchase loans, suggesting a bifurcated market.
  • While loans to all groups of borrowers increased, high-income borrowers, non-Hispanic whites, and Asians showed significantly larger increases than other groups. Denial rates for black and white-Hispanic applicants continued at rates significantly higher than denials for other groups. The big reasons for denial related to too high loan-to-value or debt-to-income ratios.
  • Looking at census tracts rather than individual borrowers, lending was weakest in low-income and minority census tracts.
  • High-cost lending (a proxy for subprime lending) remained low, at 3 percent (compared to 28 percent in 2006), but black and white-Hispanic borrowers were more likely to have higher-priced loans.

The matched data enabled Bhutta and Canner to add credit scores—missing from HMDA but scheduled to be included under the Dodd-Frank Act—to their analysis, as well as performance. Some important findings:

  • Confirming the existence of a tight “credit box,” average credit scores rose significantly from 2006 to 2010, from 701 to 728 for home-purchase loans and 682 to 772 for refinances (or “refi”). The inversion of the comparative credit scores for purchase and refi likely reflects the changing shares in origination channels, with FHA, VA, and RHS having a vastly increased portion of the home purchase market, whereas loans banks held in portfolio or sold to Fannie Mae and Freddie Mac were largely refis.
  • In 2010, 22 percent of home purchase borrowers had back-end (i.e., both housing and non-housing) debt payment to income ratios above 43 percent, the bright-line standard for a “Qualified Mortgage” (QM) under the Consumer Financial Protection Bureau’s (CFPB) new rules. However, 70 percent of that 22 percent were FHA, VA, or RHS. On September 30, the FHA released its version of the QM rule; it does not have a 43 percent debt-to-income cutoff.
  • In both 2006 and 2010, loans made by institutions subject to the CRA in their CRA assessment areas had higher credit scores and lower debt–to-income ratios than loans originated outside assessment areas. CRA assessment area loans made in 2006 to low- or moderate-income borrowers had significantly lower delinquency rates than the universe as a whole, and one-fourth the delinquency rate of higher-priced loans. Bhutta and Canner conclude that “the relatively low delinquency rate of loans encouraged by the CRA is inconsistent with the notion that the CRA was a principal driver of the mortgage and financial crisis."

Starting this year, important improvements spurred by work at the CFPB will make the data more accessible. The data on the website of the Federal Financial Institutions Examination Council are easier for researchers to retrieve and use. In addition, the CFPB’s own website now has graphs of loan originations by loan use (purchase, refinance, home improvement) and loan type (conventional, FHA, VA, RHS) for 2010, 2011, and 2012, as well as information about changes in mortgage applications and originations.

CFPB will shortly release a far more robust query tool as well as an API to help developers access HMDA data and design their own applications and visualizations. The Urban Institute’s own Data Dashboard and National Data Repository will also be updated shortly with 2012 HMDA data.

Image from Shutterstock

Share this:
Share this page via Email Share this page via Stumble Upon Share this page via Digg this Share this page via Facebook Share this page via Twitter
Filed under: Built Environment |Tags: , , ,
Add a Comment »

Leave a Reply