Why economic forecasts keep missing, and how to use them anyway

Institutions publish central projections because decisions require a number. Behind almost every one is a distribution, and the published uncertainty bands are wide enough that the central path is rarely the outcome.

Where models break

Macro models extrapolate relationships estimated on past data. They handle continuation well and discontinuity badly. Recessions, policy regime changes and supply shocks are precisely the events that break estimated relationships, which is why they are systematically under-forecast.

Read the assumptions

• The interest rate path assumed — often market pricing, not the forecaster's view.
• Energy and commodity price assumptions, usually frozen at a recent average.
• Fiscal policy assumed as legislated, ignoring likely changes.
• Population and participation assumptions, which drive medium-term output.

Consensus is not accuracy

Forecasters cluster, because being wrong alone is professionally more costly than being wrong together. A tight consensus therefore indicates agreement, not confidence, and the largest market moves occur when that consensus is displaced.