[New Paper] An introduction to conformal inference for economists

Conformal inference is an exciting new statistical tool that has emerged from computer science. It offers finite-sample prediction intervals with strong guarantees — without requiring distributional assumptions beyond exchangeability. While becoming widely adopted in machine learning, conformal methods remain fairly unknown in economics. As prediction methods become more central to econometrics, both in terms of forecasting and inference, I believe these methods will become more important and influential.