The national forecast, allocated by state

State forecasts are the national GDPcast forecast allocated by each state's industry mix plus a momentum term — they are not independently modeled from state-level data. That means every state forecast carries all of the national forecast's own uncertainty, on top of whatever error the allocation step adds.

On 4-quarter-average growth: RMSE 1.61pp, versus 1.79pp for the naive benchmark of “every state grows at the national rate” (-10%), MAE -12.5%, cross-state rank correlation 0.41. Read the map as which states are positioned to out- or under-perform and why, not as point estimates good to a tenth.

Each state's forecast splits into an industry-mix effect (how that state's industries are weighted versus the nation's) and a momentum effect (how the state has actually run versus what its mix alone would predict, recently — shrunk toward zero and capped at ±1.5pp so a few noisy quarters can't swing the forecast on their own).

GDP-weighted state growth is reconciled to equal the national GDPcast forecast exactly — the state map is a decomposition of the national call, not a separate, potentially-inconsistent set of 51 forecasts.

Industry shares and momentum are frozen from BEA GDP-by- state data (currently BEA GDP by state through 2026Q1), refreshed quarterly as BEA publishes its next release — not recomputed on every GDPcast run.