Does the effect of the stimulus depend on which projects it funds or how big it is?
Game-theoretic analysis suggests that:
- The current crisis can be modeled as a coordination game.
- The government is uniquely capable of causing a transition from the suboptimal equilibrium of the game to the optimal equilibrium.
- Its ability to do so is dependent only on the fact that the stimulus does anything to cause more private spending and is large enough to signal credible commitment.
My friend Oggie writes:
"...if you want to get out of crisis you need to realign supply and demand. Simply spending money on projects will not do that. Even if these projects are beneficial they might still not to the job of setting up a more sustainable real economy, which produces what consumers want. Stimulus might employ people, but what happens when the government stops spending money? Presumably those employed by the government projects will have to go and find another job, and the firms that were getting revenue from the government will have to scale back on their hiring. People get fired, consumption drops and businesses scale back on hiring, which leads to even more job cuts...the same problem we have now."
I agree that in the long run, we need the economy to be broadly market-determined, as Oggie suggests; I don't think very many politicians are thinking of the stimulus package as designed to create long-term government jobs. If the only effect of the stimulus was to temporarily change the number of people employed by the government and the revenue generated by companies, then it would clearly be a distorting proposition and potentially slow our recovery.
I think the most legitimate rationale for the stimulus, however, lies in a model that conceptualizes the problem somewhat differently. Nate Silver suggests that some corporations--obviously, I can't give or find actual numbers, since the relevant data is all confidential--are essentially waiting for the other guy to move first.
Here's why: if I go ahead with a project in this economy, the likely result is worse than if the project were undertaken in a good economy (think, for example, of launching a new retail product: however difficult it may be in a normal economy, people are especially reluctant to buy in a bad economy). Thus, I am willing to take fewer risks and pay less for the opportunity. If somehow we could all coordinate to move together, then the economy would be good--employment would go up, incomes would rise, demand would rise, and revenues would follow, allowing us to grow.
Formally, this situation is modeled in game theory as the stag hunt. Basically, I prefer the safe option if I believe you are too, and I prefer coordinating with you to that outcome, but if I take a risk while you play it safe, I'm in trouble.
With two people playing this game, it's relatively easy to coordinate--since both of us have an incentive to do so, we can just say to each other "let's coordinate" and, when we are, neither of us would rather not. This equilibrium is self-enforcing.
What if there are 10 people playing? 100? 300,000,000? What if the "cooperate" outcome only occurs when more than some critical fraction of the society decides to cooperate?
Then we need leadership. This kind of socially agreed-upon decision is essentially what Thomas Schelling calls a focal point. That is, in a situation with multiple equilibria (in the above game, the both-cooperate and both-play-it-safe outcomes are equilibria, because once they are picked, nobody would unilaterally prefer a different strategy), anything that makes one more salient than the others can help to coordinate our choices.
Therefore, I see the role of the government in the current crisis to one-up the oracles of old: not only can the government (and the President, in particular) say "okay, everybody, we're all going to move together on this"--as, incidentally, President Bush did after 9/11--but the government can actually take the first step, assuming all the risk of doing so.
So let's go back to our game with 10 people, and say 5 people have to cooperate in order for everybody to get some common benefits. Given the size of the government, it can stand in for 3 people. As you can imagine, it's much easier to reach 5 people cooperating if 3 people already are, since it reduces the risk of being left stranded.
The real point of all this is that the details of the size and composition of the stimulus package may not actually make that much difference. Sure, we would prefer the money be spent on the maximally-beneficial projects, but as long as the government is actually paying people to do things, it is diminishing the dangers faced by other organizations who are considering taking risks. The key then becomes whether the package is large enough to be a credible and useful step forward. Tax cuts, under this reading, may not be as stimulative precisely because they keep the responsibility for coordinating in the hands of massively distributed agents, rather than one focal-point-inducing central organization.
Is ~5% of GDP enough to be credible and useful? It appears we will see.
[cross-posted at Oeconomica.blogspot.com, the blog of the economics research group Oeconomica]