|Panel Data and Experimental Design|
with , : w26250
How should researchers design panel data experiments? We analytically derive the variance of panel estimators, informing power calculations in panel data settings. We generalize Frison and Pocock (1992) to fully arbitrary error structures, thereby extending McKenzie (2012) to allow for non-constant serial correlation. Using Monte Carlo simulations and real world panel data, we demonstrate that failing to account for arbitrary serial correlation ex ante yields experiments that are incorrectly powered under proper inference. By contrast, our “serial-correlation-robust” power calculations achieve correctly powered experiments in both simulated and real data. We discuss the implications of these results, and introduce a new software package to facilitate proper power calculations in practice.
Published: Fiona Burlig & Louis Preonas & Matt Woerman, 2020. "Panel data and experimental design," Journal of Development Economics, .
|Machine Learning from Schools about Energy Efficiency|
with , , , : w23908
In the United States, consumers invest billions of dollars annually in energy efficiency, often on the assumption that these investments will pay for themselves via future energy cost reductions. We study energy efficiency upgrades in K-12 schools in California. We develop and implement a novel machine learning approach for estimating treatment effects using high-frequency panel data, and demonstrate that this method outperforms standard panel fixed effects approaches. We find that energy efficiency upgrades reduce electricity consumption by 3 percent, but that these reductions total only 24 percent of ex ante expected savings. HVAC and lighting upgrades perform better, but still deliver less than half of what was expected. Finally, beyond location, school characteristics that are readily ...