🤖 AI Summary
This study addresses a critical limitation of fixed-effects causal forests in estimating conditional average treatment effects (CATE): the systematic attenuation of effect heterogeneity due to averaging within leaf nodes, which underestimates the true variability of treatment effects. The authors demonstrate that this bias stems from the leaf-node averaging mechanism and formally characterize its dependence on signal-to-noise ratio, panel size, and dimensionality. To correct this attenuation, they propose a self-contained, out-of-bag cross-fitting debiasing procedure that recovers the true CATE distribution by rescaling the estimated slopes. The method integrates causal forests, within-transformation for fixed effects, and optimal linear prediction, and is implemented in the `causalfe` Python package. Simulations show that the proposed correction reduces CATE mean squared error by 25–42% compared to standard recentering, and an empirical application to county-level minimum wage data successfully restores previously compressed effect heterogeneity.
📝 Abstract
Causal forests that estimate conditional average treatment effects by averaging honest leaf-level effects across trees are widely used in fixed-effects panel settings. We show that this averaging systematically attenuates the estimated heterogeneity: the raw prediction behaves like a + b*tau(x) with slope b < 1, so the spread of the CATEs is compressed toward the average effect, and the additive recentering used to report an unbiased average treatment effect does not fix it. Benchmarking against a similarity-weight generalized random forest on the same within-transformed signal, we find both estimators attenuate but the leaf-averaging construction attenuates materially more. We characterize how b moves with the design, worsening with lower signal-to-noise, smaller panels, and higher dimension; this diagnosis is our main contribution. As a remedy we adapt the best-linear-predictor calibration of Chernozhukov et al., estimating the de-attenuation slope out-of-bag so that it is self-contained within the observational panel and asymptotically inert under a homogeneous effect. In simulations the correction cuts CATE mean-squared error by 25-42% relative to the recentering default; on a standard county minimum-wage panel the attenuation is present but mild and the correction restores the imposed spread. We ship the method in the causalfe Python package.