Economics Seminar - "Regulatory Frictions, Prescribing Behavior, and Unintended Consequences: Evidence from Gabapentin Policies"

Regulations, information constraints, and administrative burdens are key drivers of provider behavior; we use recent state-level policies targeting gabapentin as a natural experiment to study these mechanisms in prescribing decisions. Gabapentin, an increasingly misused medication with rapidly expanding off-label use, has been subject to two distinct regulatory interventions: Drug-of-Concern (DOC) designation, which expands PDMP reporting without constraining prescribing, and Schedule V classification, which imposes binding prescribing and documentation requirements. Using a novel dataset linking MarketScan claims (2016–2024) with CDC multiple-cause-of-death records, and a stacked difference-in-differences framework, we document four main findings. First, DOC designation has no detectable effect on gabapentin prescribing. Second, Schedule V classification leads to a sustained decline in prescribing, reducing gabapentin prescribing by 18 to 25 percent (from a pre-policy baseline of 12.2 patients per 1,000) within eight quarters, driven by reduced continuation among established users. Third, we find little evidence of harmful substitution: opioid prescribing does not increase, co-prescribing with opioids declines, and we find no evidence of substitution toward pregabalin. Fourth, Schedule V classification is associated with a meaningful reduction in gabapentin-involved mortality, including opioid-co-involved overdose deaths, with no corresponding change following DOC designation. Mechanism tests exploiting variation in PDMP mandates and patient histories show that neither salience nor information effects generate measurable changes in prescribing; instead, the behavioral response is concentrated in settings where Schedule V classification raises administrative hassle costs. Collectively, the results demonstrate that only binding regulatory frictions meaningfully reduce gabapentin use and associated harms, while monitoring-only policies are largely ineffective.