The Keynesian Multiplier reconsidered