Unknown cause of missing Lending Club loan descriptions after March 2014

Ascertain why the Lending Club dataset no longer contains the desc (loan description) field after March 2014, which necessitated restricting the study period to June 2007–March 2014.

Background

The paper relies on the desc variable (borrower-provided loan descriptions) to compute BERT-based risk scores. The authors note that descriptions are missing beyond March 2014, forcing them to limit the dataset to earlier years.

The reason for this data unavailability is explicitly stated as unknown, and resolving it would clarify data coverage and potential replication or extension of the analysis to later periods.

References

For unknown reasons, we noticed that the descriptions were no longer available as of March 2014, so we further filtered the dataset considering only the loans from the beginning (June 2007) until that month.

— Credit Risk Meets Large Language Models: Building a Risk Indicator from Loan Descriptions in P2P Lending (2401.16458 - Sanz-Guerrero et al., 29 Jan 2024) in Section 4 (Description of the dataset)