Panel Paper: An Evaluation of State Housing Finance Agency Loan Performance

Saturday, November 4, 2017
Wright (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Stephanie Moulton1, Matthew Record1 and Erik Anders Hembre2, (1)The Ohio State University, (2)University of Illinois, Chicago

Between the late 1970s and 2010, state housing finance agencies (HFAs) across the country financed $260 billion in mortgages. These organizations funded home purchases for more than 2.9 million Low-to-Moderate Income (LMI) households using the sale of mortgage revenue bonds (MRBs) (National Council of State Housing Agencies, 2011). In recent decades, HFAs have played a key role in providing lower-cost mortgages for credit-worthy but underserved populations. While historically low interest rates over the past few years have reduced the sale of MRBs, HFAs have leveraged alternative financing so they may continue to provide mortgages to LMI borrowers. Over the past decade, a sizeable proportion of HFA loans have been structured as conventional mortgages guaranteed by Fannie Mae (Moulton and Quercia, 2013).

Preliminary evidence suggests that HFAs are effective in their single family lending practices, originating mortgages that compare favorably to non-HFA mortgages in terms of loan performance (Moulton and Quercia, 2013). However, these preliminary studies are based on survey data or a handful of state HFA programs. The present research addresses this gap by utilizing a database of nearly 120,000 HFA loans originated from 2005-2014 purchased by Fannie Mae, as well as a comparison group of more than 500,000 non-HFA, first-time homebuyer loans purchased by the same institution. Utilizing a Coarsened Exact Matching strategy, treated units are matched on a number of key criteria including geographic proximity, the year of loan origination, borrower characteristics such as FICO scores and characteristics of the loan itself such as combined loan-to-value ratio. Having established a suitable control group, analysis is conducting utilizing hazard models that measure the length of time to outcomes such as default or prepayment of the loan as well as state fixed effects models, which measure overall propensity to default.

Initial findings suggest that during the period leading up to the beginning of the recent housing crisis (2005-2008) HFA loans out-perform otherwise similar non-HFA loans among first-time homebuyers, with generally longer periods of time to default and a generally diminished propensity to ever go into default overall. However, the observed association disappears in the period during and following the crash (2009-2014) suggesting that tighter underwriting standards adopted by non-HFA financial institutions following the housing bust may have closed the gap between HFA and non-HFA loan performance in recent years.