Hi, I'm Larry Nitz. I'm a professor of Political Science at the University of Hawaii. My whole professional life, I've been engaged in this boundary between politics and economics. Our inspiration has always been, in many cases, for the market, Adam Smith. But we mostly forget that he wrote another book, other than The Wealth of Nations, called Moral Sentiment. What Adam did is, wrote a manual for the market, and he wrote a manual for the early version of the Uniform Commercial Code. He was concerned with transactions being fair, with counts being complete, with contracts being executable. The Consumer Financial Protection Board was designed to help people in the United States in their transactions with financial organizations to protect them against ill-conceived transactions, transactions that couldn't be enforced, less than honest behavior, or just nastiness. In its lifetime, the Consumer Financial Protection Board collected on the order about two and a half million consumer complaints. Four hundred thousand of these were complaints about mortgage lenders. Forty-eight hundred of these about reverse mortgage lenders. The CFPB data has ceased. This may be the last opportunity to examine contemporary complaint cases because they're going to be older and older over time.

In light of past subprime mortgage scams, borrower mistreatment may be discriminatory against minorities and poorer folks, but the CFPB has no ethnicity or income status in complaint records. By zip code, we can match folks to their census district via the zip code tabulation area statistics. We can raise questions about the likelihood of reverse mortgages versus forward mortgages by the neighborhood economy. The nature of complaints by reverse and forward mortgage borrowers...

Issue differences for reverse and forward mortgage in high minority areas. Issue differences in high-income or property value areas. The first quick check is if we do a logistic regression on the probability of having a reverse mortgage from the total population of 400,000 mortgage complaints, we can include the proportion Black by zip code, the proportion Hispanic by Zip code, the adjusted Housing value by zip code, and the adjusted median income by zip code. Of these, the two ethnicity variables predict absolutely nothing. High property values and high income are predictors of reverse mortgages. Reverse mortgages are not a target device to hit the poor, but why are they a problem? One way to catch this is to look at the major issues that are reported, and the CFPB encodes these pretty nicely, people can pick them off our list.

For forward mortgages, people have the most complaints about paying, about renegotiating, about rates. For reverse mortgages, the big one, the one that counts, is people have trouble paying it off. Now, what? Why should you have trouble paying off a mortgage? You're just saying, "Let me bring in the money and I will pay off the mortgage that my mother borrowed." But that seems not to be quite the way it works. We know at this point, none of the ethnic composition or economic composition variables distinguish much between reverse and forward normal mortgage-issued complaints. If we turn to the structure of complaint narratives. Without spending a whole lot of time on the scores and the word clusters, for forward and reverse mortgages, we get different narratives. For forward mortgages, regular mortgages, the first cluster contains almost exclusively the names of lenders. It's not flattering. The second cluster seems to reflect issues with the accounts and escrow services. That is, was my payment recorded and did they collect money for the taxes and the insurance? Cluster 3 touches on communications with the lender or the servicing agent.

For reverse mortgages, the first cluster mentions both lenders and transactions, that is, monetary draws, taking money out. The second cluster expresses a level of dissatisfaction and concerns about insurance, that is, whether the insurance is recognized and whether the mortgage company placed insurance with their own carrier. Cluster 3 seems to reflect dissatisfaction communications with a lender. A slightly different crop of complaints for specification of a larger number of clusters might produce different but no more confident or interpretable clusterings.

We can look at the word cloud, and the word clouds are always fun. In this case, they look much the same. We have a set of complaints, and in these complaints, the big words are good, working, great, best, better, appropriate, satisfied. Yes, there are problems, and wrong, and bad, and unfortunately, but we don't have a really good sense of what's going on here. We can read anger and judgment into the negatively scored text. People say they're unprofessional, and their incompetency is completely unacceptable, or this is the worst company, or the conduct of NFCU and its employees during this experience uniformly unsatisfactory. That's unhappiness.

We can look at the text that is scored for sentiment. Now, let's take a few expressions from text scored between 0 and minus 50. It's not minus 70, not minus 80. It's minus 50 on the whole text, so it's moderately negative. Homeowner was not legally able or mentally capable of entering into any legal agreement. I sought to inquire about my status as a successor in interest and a non-borrowing spouse. Unfortunately, I was told I do not qualify under their program. However, after consulting an attorney, I've learned that I do qualify as a non-borrowing eligible spouse. Here we're talking about the right to live in the house afterwards, after your husband dies. My mother's house. She did a reverse mortgage with reverse mortgage solutions before she died. When she died, we reached out to reverse mortgage solutions, trying to get details and payoff information to settle the account. We were told we weren't on the account and couldn't have information. Another one, after a year of negotiating and submitting all the paperwork they required for multiple times because the file kept getting misplaced, they tell my mortgage broker that they're no longer the bank and to submit everything again to the new lender, which we do with no response. Here's somebody trying to buy a new mortgage, to put a new mortgage on mom's house and pay off the old one, and it doesn't work.

Finally, AAG authorized an additional X thousand dollars on a line of credit to my mother's reverse mortgage. The signature on file is not my mother's. This would be simple to check from AAG. My mother was in the hospital during the first authorization, would not have been able to sign and fax these forms. Now, that sounds a little bit disturbing. What we find here is that these expressions don't use traditional, emotionally charged language. These are pretty negative events. The most negative express anger and disapprobation in explicit terms. Unprofessional, nasty, all that stuff. The least negative are generally presented in neutral language but convey clear implications of loss. For closure, auction schedule. No notice to the family of foreclosure proceedings. Lender will not acknowledge family's legal or estate executive representation. Removal or exclusion of spouse from remorse mortgage documents. Required resending of documentation. Adding all these outcome expressions, schedule force to closure example, does not outweigh the influence of the unhappiness terms.

Maybe there's a way to tie this down. If we take a four-topic singular decomposition of the terms, we can get concepts like loss and blame, bureaucratic challenges, legal issues, and frustration. We map this along with the negative sentiment and positive sentiment mean scores.

We find the negative sentiment mean scores fall directly on the axis of democratic hurdles faced by families. The loss vector and legal issues are split by negative sentiment on either side. We want to use sentiments to flag good or bad reactions, but as long as the language is expressive, this is easy to do. I imagine in a dispute about a divorce, the bad reactions and the bad outcomes are all in the same language. But when language becomes intentionally measured, as illegal claims, it is hard to collect enough events, loss of home, foreclosure, forced out, to dramatically shift the sentiment tallies from bad words to bad outcomes. But we can map them together, as you've seen. It's inescapable here that surviving spouses and heirs are sorely dissipated and pointed in the reverse mortgage industry, particularly dealing with their agents. Supporters with appropriate skills, attorneys and accountants, are being boxed out of representing their clients, and reverse mortgages have, for many people, been a way to steal their home. Thank you.