Arnold Kling cleverly debunks

government spending multiplier mythology:

Brad Setser writes,

Facts are facts. The US has already proved it can raise over $1.5 trillion in a single year [in Treasury borrowing]

That is a the sort of statement that could come back and haunt someone. It is along the lines of the guy jumping out of a building from the 10th floor, passing the third floor and saying, "It's all fine so far."

It is amazing what happens when you assume that you live in a linear world. You say that the multiplier for government spending is 1.57.

Really? Over what range? Think of it this way: at which level of additional government spending would the path of U.S. real GDP be the highest?

(a) $100 billion in spending above the baseline
(b) $1 trillion in spending above the baseline
(c) $100 trillion in spending above the baseline

If you use a constant multiplier of 1.57, the right answer is (c). Yet we know that this is not the right answer. At $100 trillion in additional government spending, the United States would be operating like Zimbabwe, with similar results.

So to talk about "the" multiplier, as if it were linear, has to be wrong at some level. Is the multiplier linear over the range between $100 billion of additional spending and $1 trillion of additional spending? I think that is unlikely. Between, say $400 billion and $800 billion, is the incremental multiplier still in a range between 1 and 2? I worry that it is much lower. I worry that it turns negative somewhere in that range.