A new NBER working paper by Lunsford and West tests six forecasting models across ten long‑run, cross‑country macro variables and finds that very simple models (AR(1), frequency‑domain approaches, or random walk where appropriate) deliver reasonably well‑calibrated 10– and 25‑year forecast distributions. In short: for many macro variables, long‑horizon uncertainty is smaller and more predictable than commonly feared.
— If long‑run macro outcomes are more predictable, policymakers and voters should temper panic responses to short‑run shocks and refocus debates on structural reforms and coordination problems.
Tyler Cowen
2026.03.03
100% relevant
NBER working paper by Kurt G. Lunsford and Kenneth D. West testing models on long‑run annual cross‑country data and Tyler Cowen’s public note (plus a GPT 'plainspeak' translation) highlighting the result.
← Back to All Ideas