Does College Level the Playing Field? The Homeownership and Mortgage - PowerPoint PPT Presentation

Does College Level the Playing Field? The Homeownership and Mortgage Experience Carlos Garriga Assistant Vice President, Research Division Lowell Ricketts Senior Analyst, Center for Household Financial Stability Don Schlagenhauf Chief Economist, Center for Household Financial Stability Federal Reserve Bank of St. Louis www.stlouisfed.org/hfs *These are my own views, and not necessarily the views of the Federal Reserve Bank of St. Louis, Federal Reserve System, or the Board of Governors

Motivation  Households headed by college-educated Black or Hispanic/Latino individuals fared worse in the 2007 – 2013 period than Whites and Asians, as well as their peers who didn’t earn a degree.  During this same period, the U.S. suffered the worst housing crisis in modern memory and the second worst recession.  How much of the divergent wealth outcomes can be explained by different mortgage borrowing and homeownership experiences during this turbulent time?

Takeaways  Compared to whites, minority groups held a greater concentration of variable rate loans, including interest-only ARMs.  These loans were associated with greater default rates.  Minority groups had higher default rates across all products, including fixed rate loans.  Individuals with higher incomes had higher ownership rates of riskier loans (ARMs, I.O. ARMs).  Assuming college-educated individuals occupy higher income quartiles; defaults were concentrated among college-educated Hispanic/Latinos.

Overview  Introduction  Data, Sample Choice, Summary Statistics  Loan Products and Default Rates, By Race/Ethnicity  Results by Income Distribution Within Racial/Ethnic Groups  Conclusion

Introduction  All racial and ethnic groups embraced the dream of homeownership during the run-up to the housing crash.  For many college graduates, purchasing a home of their own was the logical next step.  With few accumulated assets, their financial stability and hopes of wealth accumulation were heavily reliant on this investment in housing.

Homeownership Peaks Close to Our Sample Date Homeownership Rates by Race and Ethnicity of Householder, 1994 to Present Percent 80 75 70 65 60 55 50 45 Total 40 White, Non-Hispanic Black/African-American 35 Hispanic, any Race 30 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: Current Population Survey/Housing Vacancy Survey, Series H-111, U.S. Census Bureau.

Data  Merged Confidential Home Mortgage Disclosure Act (HMDA) with a match of loans in the McDash dataset.  Time series data by individual loans with detailed borrower information best suited to assess mortgage experience.  Total match includes over 89 million loans originated between 1992 and 2014.  Random subsample of loans originating in June 2005 for white, non-Hispanic borrowers and larger sample for minority groups.

Data  Advantages:  Direct measurement of demographics (reported at loan application) avoids identification issues.  Monthly frequency gives dynamic look at loans at origination, termination, and everything in between.  Direct measurement of income at the time of origination.  Disadvantages:  No direct measurement of educational attainment.  Income is static over time, doesn’t capture job loss.

Data  Our analysis focuses on four groups: Black/African-Americans, Asians; Hispanic/Latino, any race; White, not Hispanic.  We have unique loans distributed as follows:  Asian: 4,680  Black/African-American: 7,970  Hispanic/Latino, any Race: 11,865  White, Non-Hispanic: 9,024

Lower FICO Scores, Higher Rates, Lower Appraisal Summary Statistics by Race/Ethnicity, March 2006 Variable Asian Black Hispanic/Latino White Median Applicant Income 97 62 68 76 Median FICO Score 729 656 685 719 Median Appraisal Amount 385 189 262 233 Median Interest Rate 5.875 6.375 6.25 5.875 Share Fixed 46.9 55.7 51.2 64.1 Share Variable 52.3 43.5 48 34.6 NOTE: All dollar amonts are in thousands.

L OAN P RODUCTS AND D EFAULT R ATES

Loan Product Definitions  Fixed-Rate Mortgage : Mortgage loan where the interest rate stays the same over the life of the loan.  Adjustable Rate Mortgage (ARM) : Mortgage loan where the interest rate periodically adjusts based on an index.  Interest-Only ARM : An adjustable-rate loan with an initial interest-only payment period. No principal is paid during this time. Monthly payments increase substantially after the period.  Balloon Mortgage : A mortgage which has a balance due at the end of its life. This “balloon” payment is usually large.

Minority Groups had More ARMs, Specialty Loans Distribution of Loan Products Percentage Share of Loans Originating in June 2005 70 Fixed 62 ARM 60 I.O. ARM I.O Period Unknown 53 52 Balloon unknown 50 50 46 40 33 32 30 29 30 22 20 16 13 12 11 10 10 4 3 3 3 3 0 Total Asian Black/African-American Hispanic/Latino, any Race White, Non-Hispanic

Highest Default Rate for Interest-Only ARMs Default Rates, By Product Type Percentage Share of Loans Originating in June 2005 18.1 I.O. ARM I.O Period Unknown 15.1 ARM 7.5 Fixed 7.0 Balloon unknown 0 2 4 6 8 10 12 14 16 18 20

Default Rates Higher Across All Products Default Rate, by Loan Product and Racial/Ethnic Group Percentage Share of Loans by Type Originating in June 2005 30 Fixed ARM 24 25 I.O. ARM I.O Period Unknown Balloon unknown 20 20 19 18 16 15 15 14 11 11 11 10 10 9 10 8 7 7 6 5 5 5 5 0 Total Asian Black/African-American Hispanic/Latino, any Race White, Non-Hispanic

Real Estate Owned Sales are Majority of Defaults Loan Defaults, by Type Termination and Race Percentage Share of Loans Originating in June 2005, Sample Ends in June 2013 9 8.3 REO Sales 8 Liquidation to Termination 7.3 Payoff out of Foreclosure 7 6 5 4.3 3.9 3.6 4 3.3 3.3 3 1.9 2 0.8 1 0.6 0.4 0.3 0 Asian Black/African-American Hispanic/Latino, any Race White, Non-Hispanic

Underwater Loans Were Motivating Factor

R ESULTS BY I NCOME D ISTRIBUTION

Income Distribution, Rough Proxy for Education

Asian’s had Greatest Holdings of ARMs Product Type, By Income Quartiles, Asian Group Percentage Share of Loans Originating in June 2005 80 1st Quartile 2nd Quartile 70 3rd Quartile 4th Quartile 60 60 47 50 44 43 40 34 32 32 30 25 22 18 20 16 10 10 3 3 3 2 0 Fixed ARM I.O. ARM I.O Period Unknown Balloon unknown

ARM Ownership Rate Similar Across All Quartiles Product Type, By Income Quartiles, Black/African-American Group Percentage Share of Loans Originating in June 2005 80 1st Quartile 2nd Quartile 70 3rd Quartile 4th Quartile 59 60 54 52 50 45 40 34 33 32 31 30 20 14 11 9 10 4 4 4 3 3 0 Fixed ARM I.O. ARM I.O Period Unknown Balloon unknown

Pronounced Positive Relationship, Income w/ ARMs Product Type, By Income Quartiles, Hispanic/Latino Group Percentage Share of Loans Originating in June 2005 80 1st Quartile 2nd Quartile 70 64 3rd Quartile 4th Quartile 60 52 50 45 37 40 36 32 29 30 25 21 20 15 12 10 6 4 4 4 3 0 Fixed ARM I.O. ARM I.O Period Unknown Balloon unknown

Positive Relationship Holds for Whites as Well Product Type, By Income Quartiles, White, non-Hispanic Group Percentage Share of Loans Originating in June 2005 80 71 1st Quartile 2nd Quartile 68 70 3rd Quartile 4th Quartile 60 60 49 50 40 28 30 23 19 19 19 20 12 9 7 10 4 2 2 2 0 Fixed ARM I.O. ARM I.O Period Unknown Balloon unknown

Defaults for Hispanic/Latinos Among Highest Earners Distribution of Loan Defaults, by Income Quartiles within Racial Groups Percentage Share of Loans Ending in Default, Sample Ends in June 2013 40 1st Quartile 2nd Quartile 34 33 35 3rd Quartile 4th Quartile 29 30 28 27 26 25 25 23 25 22 22 22 21 21 21 21 20 15 10 5 0 Asian Black/African-American Hispanic/Latino, any Race White, Non-Hispanic

Conclusion  Risky loan products were concentrated in the upper income quartiles, which we assume encompass the college-educated.  During the crash those products had more severe default rates.  In general, default rates were higher for minority groups.  Looking at defaults by income shows a concentration of defaults among Hispanics/Latinos who earn the most.

Conclusion  Given that our sample only includes individuals who have applied and been approved for a loan we likely don’t have the true bottom of the income distribution.  This suggests that the default rates within the lower income quartiles also capture the experience of some college educated individuals.

Further Research  Benchmark our income distribution with that of Survey of Consumer Finances and Census, who is missing and what can we say about educational coverage?  Within the Interest Only ARMs and ARMs in general, what idiosyncratic characteristics are predictors of default?  What if we look at geographic flashpoints of the housing crisis (NV, CA, FL) and the minority communities there. Are there different trends within these local markets?

For More Information… Center for Household Financial Stability stlouisfed.org/hfs In the Balance Research Essays stlouisfed.org/publications/in-the-balance FRED Economic Data research.stlouisfed.org/fred2