By Request: Stimulus Size

Mickslam wants to know “The size of the stimulus necessary to get us out of the recession and avoid the depression.”

So do I! The answer is that it’s hard to say. Paul Krugman’s post on stimulus arithmetic illustrates the issues. But as a starting point, let’s pull out of the air the estimate that absent stimulus the unemployment rate will wind up at around 9 percent. That’s about 4 percent points over what we now think is the Non-Accelerating Inflation Rate of Unemployment (NAIRU).

Now we have to think about Okun’s Law which relates increases in GDP to decreases in unemployment. Since this isn’t a real law, people are a bit fuzzy on the math. But the consensus is that to generate a one percent decrease in unemployment, you need a 2–3 percent increase in GDP. That means $300-$450 billion per percentage point or $1.2–$1.8 trillion in additional GDP.

Next is the issue of the “multiplier effect” which relates how many dollars of additional GDP you get for each additional dollar of stimulus expenditure. These are the estimates all the cool kids are using:


The catch here is that when you’re dealing with giant numbers, these multiplier estimates are going to break down. Food stamps gives you great bang for your buck. But just because $1 billion in increased food stamp spending might generate $1.7 billion in GDP doesn’t mean we could spend $1 trillion in extra food stamps and generate $1.7 trillion in extra GDP—there are only so many poor people and they can only eat so much! Much the same is true with infrastructure spending. The general idea of spending-side stimulus is to put idle assets to work doing things. We have a fair number of idle people. But to build new infrastructure you machines and so forth and we only have so many on hand. And of course you need the right people—the guys who got laid off from Lehman Brothers aren’t the engineers we need to build the Purple Line.

And of course all this is based on slightly fuzzy math about the NAIRU and very fuzzy math about likely employment trajectories. Long story short, whatever topline estimate you could do would be pretty uncertain and would run into trouble when you started thinking about implementing it in a micro sense. To make a long story short, the correct answer is a big number but the real limits probably lie in thick in the weeds rather than up in the clouds in a way that makes calculations very difficult. And then on top of all that you need to consider the international situation—will our stimulus money “leak” out of the economy in the form of a trade deficit? Will giant stimulus make the currency much weaker and alter trade flows? How much stimulus will China offer? Germany? In principle, the big surplus countries—China, Japan, Germany, Switzerland, Norway, etc.—should be doing more lifting relative to the size of their economies than should the deficit countries. But I don’t think that’s how it’s going to work in practice.