Last month I reported on a study conducted by Clarity Services, Inc., of a very large dataset of storefront payday loans and how that study revealed flaws in the statistical analyses published by the CFPB to justify its proposed rule on small dollar lending. Among the big takeaways: (a) the CFPB's 12-month study period is too short to capture the full cycle of use of a payday customer, and (b) the CFPB's use of a single-month static pool for study subjects severely over-weights the experience of heavy users of the product.
The context of the study, and of the CFPB's rulemaking, is the CFPB hypothesis that too many payday borrowers are caught in a "debt trap" consisting of a series of rollovers or rapid re-borrowings (the CFPB calls these "sequences") in which the "fees eclipse the loan amount." At the median fee of $15/$100 per pay period, a sequence of more than 6 loans would constitute "harm" under this standard.
In March Clarity published a new analysis designed to avoid the flaws in the CPFB approach, based on the same large dataset. The new study, A Balanced View of Storefront Payday Borrowing Patterns, uses a statistically valid longitudinal random sample of the same large dataset (20% of the storefront market). This article summarizes the new Clarity report.
What is a statistically valid longitudinal random sample? The study builds an accurate model of the activity of borrowers as they come and go in the data set over 3.5 years, thereby avoiding the limitations of looking at the activity of a group drawn from a single month. The sample maintains a constant count of 1,000 active borrowers over a 3.5 year sampling period, observing the behavior of the sample over a total of 4.5 years (one year past the end of the sampling period). Each time an original borrower permanently leaves the product, a replacement is added and followed.
The characteristics of the resulting sample are themselves revealing. Over the 3.5 year period, 302 borrowers are "persistent." They are continuously in the sample - not necessarily using the product every single month but visible using it periodically from the first month through some point after the end of the sampling period 3.5 years later. By simple arithmetic, 698 original borrowers drop out and are replaced. Most important, 1,211 replacement borrowers (including replacements of replacements) are needed to maintain a constant population of 1,000 borrowers who are still using the product. In other words, viewed over time, there are many borrowers who come into the product, use it for a relatively short period, and then exit forever. They number nearly four times the population of heavy users who stay in the product for 3.5 years.
Replacement borrowers are much lighter users than the persistent users who made up 30% of the original sample (which was the CFPB-defined sample). The average sequence of loans for replacement borrowers lasts 5 loans (below the six loan-threshold for "harm"). Eighty percent of replacement borrower loan sequences are less than six loans.
Turning to overall results for all types of borrowers in the sample, 49.8% of borrowers never have a loan sequence longer than six loans, over 4.5 years. Of the 50.2% of borrowers who do have one or more "harmful" sequences, the vast majority of other loan sequences (other times they use the product) involve fewer than six loans.
What does all this mean? The CFPB is legally required to balance its desire to reduce the "harm" of "debt traps" against the alternative "harm" of loss of access to the product that may result from its regulatory intervention. The current proposal imposes a very high price in terms of loss of access, eliminating 60-70% of all loans and quite possibly the entire industry. The new Clarity study shows, however, that half of all borrowers are never "harmed" by the product, and those who may be occasionally "harmed" also use the product in a "non-harmful" way more than half the time. Thus, if the CPFB is protecting consumers from "harm" while maintaining access to "non-harmful" products, it must use a much more surgical intervention than the current proposal to avoid harming more people than it helps.
This group is in debt for a payday loan, on average, sixty percent of the time. No wonder that CFPB studies that focus on this group find "debt traps."