![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
The financial health of the mortgage insurance industry is vital for the flow of mortgage credit to low-wealth borrowers. Private mortgage insurance competes with insurance offered through the federal government, particularly the Federal Housing Administration. This article employs a Heckman selection model and a database of mortgage insurance applications to examine the impact of capital reserves and federal competition on the credit decisions of private mortgage insurance companies while accounting for applicants’ decisions to seek private insurance. I find private insurers became more likely to deny applications as their capital ratios fell during the Great Recession. However, loans eligible for insurance through the Federal Housing Administration were less likely to be denied than higher loan amounts were. These findings are important for understanding how the conventional mortgage market, including government-sponsored enterprises, functions through a severe housing crisis.
Acknowledgments
The opinions expressed are those of the author, and do not necessarily reflect the policies of the Department of Housing and Urban Development or the Administration. I would like to thank Todd BenDor, Daniel Immergluck, Joshua Miller, Roberto Quercia, William Reeder, Carolina Reid, William Rohe, and Kurt Usowski for providing constructive criticism and suggestions.
Disclosure Statement
No potential conflict of interest was reported by the author.
Supplementary Material
Supplemental data for this article can be accessed at: https://doi.org/10.1080/10511482.2019.1645036.
Notes
1. The Department of Agriculture had provided home loans directly to farm workers since the Bankhead–Jones Farm Tenant Act (Pub. L. 75-210, July 22, 1937), with the Farmers Home Administration created by the Housing Act of 1949 (Pub. L. 171-81, July 15, 1949).
2. Congress also considered creating a Home Mortgage Guaranty Corporation under the Federal Home Loan Bank Board to provide partial mortgage insurance to conventional mortgages but failed to pass the legislation (Title VII of the Housing Act of 1958, S. 4035, 85th Cong. (1958)).
3. FHA typically provides coverage for the full loan amount instead of the partial coverage common among PMI companies. The first actuarial review of the Mutual Mortgage Insurance Fund notes, “[T]he difference between the 20% per loan risk taken by MICs versus the 100% per loan risk taken by FHA also renders use of the [PMI companies’] 4% capital-to-risk requirement less meaningful to FHA” (Price Waterhouse, Citation1990, p. 9). The Rural Housing Service guarantees 90% of the loan amount, whereas the VA guarantees up to 50%, depending on the loan amount.
4. Extraordinary dividends are generally defined as distributions that, combined with any other distributions made within the previous year, exceed 10% of the insurer’s statutory surplus or the statutory net income during the previous calendar year.
5. Four of the mortgage insurers settled with CFPB, agreeing to pay $15 million in civil penalties. In addition, the settlement requires the insurers not to enter any new captive reinsurance arrangements without the approval of CFPB for a period of 10 years.
6. As Veblen (Citation1904) notes, “[S]ince an advance of credit rests on collateral as expressed in terms of value, an enhanced value of the property affords a basis for a further extension of credit, and so on…. This cumulative extension of credit through the enhancement of prices goes on, if otherwise undisturbed, so long as no adverse price phenomenon obtrudes itself with sufficient force to convict this cumulative enhancement of capitalized values of imbecility.”
7. I use the definition of metropolitan statistical areas published by the Office of Management and Budget in November 2004.
8. The median length of time between application and action for PMI was 26 days, with an average of 48.5 days. Less than 9% of applications exceeded 180 days. The median for FHA-insured loan applications was 40 days, with an average of 53.7 days and less than 3% exceeding 180 days.
9. Withdrawn applications may be hidden denials; however, Munnell et al. (Citation1996, p. 40) estimate a multinomial model with withdrawn applications as an alternative outcome and find that withdrawals do not look like denials. They conclude that the data provide “little evidence to support modeling withdrawals as a lender, rather than a borrower, choice.”
10. In a previous version of the analysis (not shown), I created geographic subsamples of loan applications to use census tract fixed effects. I created a 10% national sample, a sample of bubble housing markets, and a sample of rust belt housing markets. The results are qualitatively similar to those presented here.
11. I also tested using the price index for very high- and medium-risk products, but found high-risk price had the strongest impact on selection of PMI.
12. Imputation is used for nine of the 24 quarters for Republic and six of the 22 quarters for CMG. All imputed capital ratios were for mid-year quarters before 2009. Further, because CMG was owned by PMI, when PMI went into runoff mode it stopped reporting not only its own capital ratios but CMG’s as well, even though CMG continued to endorse new loans. I do not extrapolate CMG’s capital ratios beyond that last observed figure (2011Q2).
13. The conforming loan limits themselves were similarly changed by ESA and Housing and Economic Recovery Act (HERA); however, the conforming loan limits under ESA were allowed to expire at the end of September 2011, creating a period of 27 months in which the FHA limit exceeded the conforming limit in some high cost areas.
14. For more information, see https://www.consumerfinance.gov/policy-compliance/guidance/implementation-guidance/hmda-implementation/
15. The average marginal effect is the average of the differences in estimated probabilities for each observation because of a 1-unit change in an explanatory variable (as opposed to, for example, the marginal effect at the sample means).
16. As a robustness check, I separately estimate models with an interaction between year and each of the three key selection equation variables. The estimated coefficients are shown in the appendix (see ) with 95% confidence intervals. Although the magnitudes of the effects are smaller later in the study period, they are statistically significant in almost every year. The one exception is that the FHA AMIP is not statistically significant at the 5% level in 2010.
17. Appendix shows the interaction between variables of interest and year. The effect of FHA eligibility is relatively stable across years. The effect of PMI company capital ratios inverts during the study period regardless of whether company fixed effects are used; higher capital ratios are associated with a higher likelihood of denial between 2006 and 2008 but associated with a lower likelihood of denial after 2009. However, there is much less variation in capital ratios within a given year than across years, which may inhibit robust estimation of their effects.
Additional information
Notes on contributors
Kevin A. Park
Dr. Kevin A. Park is an economist in the Housing Finance Analysis Division of the U.S. Department of Housing and Urban Development. He received his doctorate in City and Regional Planning from the University of North Carolina at Chapel Hill and this work was part of his doctoral dissertation.