Across-Design Uncertainty in Short Pricing Panels: Evidence from Simulated Price Trajectories

TL;DR

The paper examines how uncertainty in short pricing panels arises from both within-trajectory noise and variation across different price paths. It finds that most estimation error stems from differences between possible price trajectories rather than within a single observed path. Using synthetic data, it shows that accounting for this across-design uncertainty improves statistical coverage when using independent units.

Why It Matters

Many real-world pricing studies rely on short panels with limited price changes, leading to underestimation of uncertainty. This work highlights that standard resampling methods fail to capture variation across potential price paths, which can severely distort inference. Designing studies with independent price trajectories improves reliability, guiding better empirical practices in economics and machine learning.

Key Findings

• The dominant source of estimation error in short pricing panels comes from variation across different possible price trajectories, not within a single observed path.
• A scaling relationship describes across-design uncertainty as a function of the number and magnitude of price moves, with error decreasing as price variation increases.
• Using a Paule-Mandel method across independently priced units significantly improves empirical coverage, demonstrating the value of independent identifying variation in experimental design.