Value-based payment programs, also known as pay-for-performance, use financial incentives to motivate providers to invest in quality and are a critical part of Medicare health care reform. This study examines the first year of the Home Health Value-Based Purchasing program, a nationally representative cluster randomized experiment implemented by the Centers for Medicare & Medicaid Services in 2016. The goal of the program is to achieve better home health care quality. Home health agencies in treatment states were rewarded or penalized based on their performance on agency-reported and non-agency-reported quality measures. The program improved agency-reported measures by approximately one percentage point, and performance gains suggest a dose-response relationship with respect to incentive size. However, the performance gains in agency-reported measures did not reflect true quality improvement. I find evidence that agencies manipulated their coding of patients and inflated their performance. Coding manipulation explains the entirety of the program's impact on agency-reported measures.