The MBA has claimed that allowing cram-downs in Ch. 13 bankruptcy proceedings could add as much as two points to mortgage rates. Adam Levitin, an associate professor of law at Georgetown, and Joshua Goodman, an economics doctoral candidate at Columbia, have authored a study that they claim proves that allowing modifications in Ch.13 bankruptcy proceedings would not result in an interest rate spike. A recent working draft of the study, as with most academic papers, is already available (dated January 28, 2008). As an ex-academic -- HW readers may or may not know that it wasn't too long ago that I was a Ph.D. student myself -- I can't help but comment. I think the authors' data -- despite to Levitin's bellowing to the contrary -- surprisingly lends credence to the idea that altering bankruptcy law to permit cram-downs will, in fact, serve to raise mortgage rates. Study 1: historical interest rates In the first study, the authors performed a series of regression-based analyses on historical interest rate data, and included a variable intended to capture the effect of 'strip-downs.' The data used was from the time period of 1988 to 1993, a time period when strip-downs/cram-downs were allowed -- the idea here being that it is possible to tease out any effect of cram-downs in historical mortgage rate data. The finding was that allowing 'strip-downs' raised the median interest rate by 11 to 15 basis points, with the latter figure representing a statistically significant effect on a lagged six-month basis. This is seen as a "small" effect by the authors, and not a sizeable change to mortgage rates. The regressions also lead the authors to conclude that allowing cram-downs did not affect the proportion of Ch. 13 filings relative to Ch. 7 -- this is an important point in the analysis that I'll get back to. Study 2: rate survey The authors pulled current mortgage rates, PMI fees and GSE delivery fees on single-family primary residences, investor properties, vacation homes, and multi-family residences in an effort to ascertain whether or not current rates diverged where cram-downs were allowed under current bankruptcy law. While rates/fees generally diverged significantly between primary residences and investor properties, the authors found no such disparity between primary residences and other property types subject to cram-down risk. Thus, the authors concluded that cram-down risk could not explain the divergence between primary residences and investor properties. Debunking the debunking Let's start with study one. I've noted in some earlier commentary that the Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 has had a dramatic impact on the bankruptcy landscape. In particular, the prevalence of Ch.13 filings has shot up dramatically since the enactment of the Act, while overall bankruptcy filings have dropped. In the paper, the authors manage to test whether strip-down rulings have a bigger effect on mortgage rates in states where Ch. 13 filings are more prevalent. What they find is statistical relevance suggesting that every 10 percentage point rise in the proportion of filers using Ch. 13 leads to an additional 17 basis point increase in mortgage rates when strip-downs are allowed. This 'prevalence effect' comes over and above the 10-15 bp rise found as a so-called main effect of allowing cram-downs. It's a finding that is sadly relegated to nearly zero discussion and cited as something to consider for "future work," when in truth it should have been the focus of much further analysis. While the authors do say they found no evidence in the historical data to suggest that allowing cram-downs drove more borrowers to file Ch.13, they completely failed to account for the so-called "means test" established under section 707(b) on the 2005 Act. That test, by its very definition, makes Ch.13 filings more prevalent. The fact that the authors were able to find strong evidence of a 'prevalence effect' at all using historical data suggests that allowing cram-downs now -- in a post-BAPCPA world -- would have a pretty dramatic impact on overall mortgage rates. The failure to consider this has to throw the entire study's conclusions into doubt. But that's not all -- let's take a peek at study two. The authors find uniform evidence of higher rates and points for investor properties, as compared to single-family primary residences; but no such divergence exists for vacation homes and multifamily residences. The conclusion, and associated footnote:
Because investor properties share the same bankruptcy modification risk as vacation homes and multifamily residences, the mortgage rate premium on investor properties cannot be attributed to bankruptcy modification risk. [footnote] It is not surprising that vacation homes have the same rates as single-family principal residences. Vacation homes reputedly have lower default rates because typically only well-heeled buyers purchase them. They do not have tenant risks such as vacancy, non-payment, or damage, and they are typically well-maintained because of the pride of ownership factor.
What is "bankruptcy modification risk," exactly? The authors don't define it explicitly, but I can assure you that investor homes and vacation homes absolutely do not face the same risk of being modified as part of a Ch.13 bankruptcy -- simply because vacation homes are less likely to default and/or be subject to a borrower bankruptcy to begin with. (It's in the authors' own footnote, for crying out loud.) I should note here that I'm not defending the 200 basis point figure generated by the MBA per se; what I've said all along is that allowing cram-downs will have a meaningful impact on mortgage rates. That could be a rise of 90 basis points, and not 200 bps; but given today's market, that's still the sort of rise that should be seen as problematic for borrowers.