I wanted to do this proper with estimates and the like but I just don’t have time. So let me sketch. I am going to leave out links as well, so I can do this pure stream of consciousness.

Garett Jones had this paper where he showed that a bunch of the jobs created by stimulus weren’t really created but taken from other employers.

Some people including Tyler Cowen have suggested that this means stimulus is not as effective as we thought because crowding out occurs through the labor market even if not through the capital markets.

This isn’t right. If you get a 50/50 unemployed to employed take up on stimulus spending then I think that gives you a multiplier around 2. That is lower than some of the estimates but not that low and it presumes that the government is no worse at finding unemployed workers than the private sector.


Well, first let me bat back what I think is the bad intuition. Suppose you had a stimulus program that provided jobs only to the unemployed and was funded through issuing of bonds which did not move the interest rate because the economy was at a zero lower bound.

Well then even ignoring the desire to lower the unemployment rate you got a completely free lunch. You extracted no capital from the economy and you extracted no labor yet, at the end you had some product – a bridge or whatnot.

That would mean that we should have stimulus out the wazoo. There are some issues with providing work that was dignified because I think that’s important for the opportunity cost of idle labor.

However, lets assume the opportunity cost was effectively zero – because we can give people dignified, skill building work – then we should stop only stop spending once we have employed every single unemployed person.

That’s because their actual product doesn’t matter. We are getting the capital and the labor for free so unless the Y has negative value, you are good to go.

It also means – if you assume the private market can do the same thing – that the multiplier is infinite. It works like this. You employ one unemployed person. They use the money to buy some stuff. The market searches another unemployed person to produce that stuff. That person then takes the money and buys more stuff which the market directs to another unemployed person and so on.

You only have to employ one person and then it whips through the entire economy and brings us back to full employment. This is because by assumption here there is absolutely no crowding out in the labor markets.  That means there is a sort of frictionless transference of stimulus.

Now this obviously isn’t how any of us think stimulus works.

That’s because implicitly we are assuming that there is crowding out in the labor market. 50/50 crowding out means that the employment transmition becomes the infinite sum of 1/2 + 1/4 + 1/8 . . . which is of course 1.

Since, you extracted no real resources to start this process off you increase GDP by 1*the initial amount of employment for a multiplier of 1.

So, seeing crowding out in that range is not inconsistent with standard multipliers. If you get 33/66 employed-unemployed hiring then you should get a 2/3 + 4/9 + 8/27 . . . which I think is 2.  That would be a big multiplier.