Why Economic Forecasts Are Almost Always Wrong (and Still Useful)
The track record of professional economic forecasting is, charitably, poor. The IMF didn't forecast the 2008 financial crisis. The Fed didn't forecast the 2021-2022 inflation surge. Almost nobody forecast COVID's economic impact in January 2020. And yet economic forecasts are produced at enormous scale and consumed voraciously. Why?
The epistemological problem
Economics involves predicting the behavior of systems with billions of interacting agents, each making decisions based on their expectations of what other agents will do. This creates a reflexivity problem: the forecast itself changes behavior, which changes outcomes. A widely publicized recession forecast increases precautionary savings, reducing consumer spending, which makes recession more likely, which partly validates the forecast through a mechanism it helped create.
This is fundamentally different from predicting a solar eclipse, where the prediction doesn't change the eclipse. Economic forecasting is trying to predict the outcome of a game while simultaneously being a player in it.
What forecasters are actually good at
Point forecasts (GDP will grow 2.3% next year) are routinely wrong by more than their stated uncertainty intervals suggest. This isn't incompetence. It's a structural feature of complex adaptive systems.
What economic analysis does better: identifying the direction of risks, articulating the conditions under which bad outcomes become more or less likely, and mapping the interdependencies that determine how shocks propagate. The 2006-2007 work by economists like Nouriel Roubini and Raghuram Rajan that identified housing market fragility was analytically correct even though their timing was off. The scenario thinking was valuable even where the point forecast wasn't.
How to use forecasts well
Treat forecasts as scenario analysis, not predictions. The value of a good economic forecast is not the central case but the distribution of outcomes it implies and the conditions it specifies for different outcomes to materialize.
Pay attention to forecaster track records and incentive structures. IMF country forecasts are systematically optimistic because the IMF needs government cooperation. Sell-side equity research is systematically optimistic because analysts need access to company management. Central bank forecasts are anchored to the path the central bank intends to pursue. Knowing the incentive tells you where to discount the forecast.