"There are two means of conducting fiscal policy, changing government spending (G), and changing taxes (T). In recessions G is increased which increases the deficit, then T is increased during booms to pay for it... Now apply this over and over, increase G in a recession, increase T to pay for it after the recovery, increase G in the next recession, then increase T again, and so on. The result is that G increases in every recession with T subsequently increased to pay for it, and government grows over time.
Now try a second strategy that switches the role of taxes and government spending. Taxes are cut during recessions, then government spending is cut during the boom to pay for it. During the next cycle, this repeats, T goes down during the recession, G is then cut to pay for it once things get better, and so on. In this case, the size of government gets smaller over time.
So, for many people... the real issue is about the role of government in the economy, some think it is too big and they support tax cuts to cure recessions, others see (correctly in my view) important unmet needs for infrastructure, health care reform, environmental initiatives, and so on, and they see government spending as the key to solving both the short-run and longer run issues."
NBER working paper whihch finds that government spending multipliers in an alternative empirically-estimated and widely-cited new Keynesian model are much smaller than in these old Keynesian models.