Close to useless

From Larry Summers sounds pretty pessimistic here by Tyler Cowen.

It has been my experience that forecasting by experts is far more fraught with error than is commonly realized though there has been increasing research and reporting on the issue in the past ten years. Here is a for instance.

The Economist had a remarkable statistic. The IMF makes forecasts for every country every April. There have been 220 instances across several decades and some number of countries where growth was positive in year T and negative in year T+1. Of those 220 instances, the IMF predicted it in April in precisely zero of those 220 instances. So the fact that there’s a sense of complacency and relative comfort should give very little comfort.

Economic turns are notoriously hard to predict, but they are among the most consequential of forecasts. To miss all 220 turns is indicative of profound lack of understanding of the deep mechanics of the macroeconomy. I don't think the IMF is alone but they might have the most embarrassing record.