Work-In-Progress
Abstract
This paper studies whether Diamond Open Access (OA) reduces p-hacking by expanding the verification environment facing researchers. We develop a simple principal–agent model in which a researcher trades off the returns to statistical significance against the expected cost of detection. Diamond OA removes paywalls and enlarges the pool of potential scrutineers, raising detection risk and generating a testable prediction: the distribution of published test statistics should exhibit less excess mass just above conventional significance thresholds. We test this prediction using the 2024 transition of Revue Économique to Diamond OA. Using more than 7,000 z-statistics from articles published between 2022 and 2025, we document clear evidence of threshold bunching in the pre-policy period that is attenuated after the reform. However, a difference-in-differences logit specification with a simulation-based synthetic control yields no statistically significant effect on threshold-crossing behavior. We interpret this result as reflecting limited post-policy observations, low statistical power, and the partial nature of the reform, which expands access without mandating data and code disclosure. Overall, the findings suggest that transparency reforms can improve reporting patterns, but their deterrent effects depend on the joint implementation of access, data openness, and verification infrastructure.
Abstract
This paper develops a continuous-time dynamic optimization model of artist behavior on digital platforms, where fanbase accumulation is both the objective and the mechanism linking artistic production and promotional effort. Within this framework, we derive the optimal intertemporal allocation of effort across a creator's lifecycle, integrating platform-mediated discovery dynamics with endogenous learning-by-doing in content creation. The model is analytically tractable and yields closed-form solutions for lifecycle strategies, platform payout functions, and regulatory benchmarks. Time-consistent behavior is shown to result in front-loaded promotion and delayed artistic production-mirroring observed creator trajectories. We prove that optimal trajectories exhibit saddle-path stability, and characterize conditions under which platforms and regulators can correct underexposure inefficiencies. The model accommodates both creative and commercial incentive structures, and offers a foundation for empirical studies of artist development, platform policy design, and cultural market regulation in the age of algorithmic discovery.
Abstract
This paper develops a continuous-time dynamic optimization model of artist behavior on digital platforms, where fanbase accumulation is both the objective and the mechanism linking artistic production and promotional effort. Within this framework, we derive the optimal intertemporal allocation of effort across a creator's lifecycle, integrating platform-mediated discovery dynamics with endogenous learning-by-doing in content creation. The model is analytically tractable and yields closed-form solutions for lifecycle strategies, platform payout functions, and regulatory benchmarks. Time-consistent behavior is shown to result in front-loaded promotion and delayed artistic production-mirroring observed creator trajectories. We prove that optimal trajectories exhibit saddle-path stability, and characterize conditions under which platforms and regulators can correct underexposure inefficiencies. The model accommodates both creative and commercial incentive structures, and offers a foundation for empirical studies of artist development, platform policy design, and cultural market regulation in the age of algorithmic discovery.