Dynamic stochastic general equilibrium models

I would think that if there was one thing we learned about economics over the past 30 years, it is that these macro models are about as predictive as a moving average. What gives?