What is the point of signing on the dotted line? In days of yore (i.e., before the E-Signature Act), signing your name with a pen was thought to caution, evidence, and channel promissory behavior. But of late, the dotted line has gotten a bad rap. In April 2018, one of the last domains of contractually-related autographs—credit cards—gave up the ghost. It seems likely that the next generation of contracting parties will never sign a physical contract. Sic transit gloria Montblanc.
Apart from pen manufacturers, should anyone care about the loss? As it turns out, there’s a body of scholarship that shows that being present at a contract’s inception—and personally marking your assent—makes later breach less likely. Several recent experimental studies have found that signing a contract has meaning—it induces caution and retards promise-breaking. Now, in an interesting draft paper, Joshua Mitts has shown that borrowers who do not personally attend their mortgage closing are much (40%!) more likely to default than buyers who are in the room where it happened. That’s true even though borrowers who skip the closing and use a power of attorney (POA) to close, are no more likely to initially show signs of financial distress.
Mitts’ first herculean task in the draft paper is to amass a dataset capable of making the claim and describe that process without benumbing the reader. He does so by combining multiple large, unwieldy databases from New York City and the Federal Government. Some of those databases must be linked through probabilistic matching (because they lacked personally identifiable information), some require OCR and pattern recognition to extract terms from mortgage contracts, while others must be hand-inspected. The result is an excess of a million mortgage/property/borrower records, mostly from the mid to late aughts. (I will admit to being slightly concerned by how easy Mitts makes it seem to match shrouded borrower data to properties.)
The paper’s central analytical challenge is to convince the reader that the sort of people who use a POA at their closings aren’t also those who breach their contracts for other reasons. Given that these are observational data, it’s not a problem that admits of an easy solution. His response is to triangulate from linear regression results: borrowers with POAs aren’t statistically more likely to be house-flippers, aren’t frequent-fliers, don’t have meaningfully different demographic characteristics, aren’t taking on different sorts of loans, aren’t more savvy, etc. (In a revision—which Mitts is working on—he’s trying to deal with the worry that POA borrowers might be more likely to be old/ill, which could be driving defaults and could be uncorrelated to other measures of wealth.)
Having chased down the selection story on borrower characteristics, Mitts interprets the results as being about selection on enhanced commitment to promise-keeping. He does so largely through an estimation showing that loans which remain with their originators are less likely to be breached—here, the POA effect “disappears entirely.” He concludes that “this evidence suggests that repeated contact between lender and borrower enhances promise-keeping, which is consistent with the hypothesis that personal promising leads to a greater sense of personal responsibility and a more salient commitment.” (P. 27.)
This is plausible, and it is consistent with experimental evidence from the contract literature (which Mitts could better use to motivate his story) that assigned contracts/mortgages are more likely to be breached. However, Mitts might have done more to try to tease out the ongoing reputational/relational story with one about commitment at the moment of the bargain. If it’s true that there is no POA effect when you control for the borrower’s ongoing relationship (if one exists), how should we think about the results?
One hypothesis (building on the signature work above) would be to try to disentangle signing a contract from the closing ceremony. The data Mitts has gathered contains some mortgages that were digitally signed and others than were signed with pens: could he test whether signing (versus clicking) has an effect? If so, then I think we could feel more comfortable with the interpretation he advanced.
In either event, Mitts is right to identify the important real-world consequences that flow from his results. This is the sort of paper that will daunt contract professors who thought that empirical work in our field required nothing more than looking at a few hundred EDGAR-coded contracts, or reading some number of appellate cases and discerning patterns. But it’s a major project that shows the continuing vitality of contractual ceremonies. I like it lots!