Casual evidence indicates that government spending during recessions (a pillar of Keynesianism) boosts GDP. But we don't have a satisfying macroeconomic model that explains why.