Driving and a Nobel Prize

Two ideas are the centerpiece of this year's Nobel Prize in economics: time inconsistency and real business cycle models. There are a lot of discussions of the policy implications of these ideas, but few of their wider import. Let me show you the beauty of real business cycles here with a non-economic example.

Real business cycles starts out with the following premise: what if the optimal choices of decision makers lead to outcomes that appear to be bad. Is this possible?

There are lots of examples of this, but let's take highway driving. Do you drive optimally? This is not saying do you drive without accidents, but rather do you make the best decisions you can given the circumstances? Let me give you a clue ... the answer is yes. You may quibble, but the point is that if you don't make the best decisions then you must be putting the lives of other people in danger with your 2-ton sub-optimally driven vehicle. Most of you won't admit to that, so let's go back to the first choice that you drive optimally.

But, if you drive optimally, what about other drivers? I know, we all think they're stupid, but can that really be the case? Probably not, they are driving optimally too.

What does all this have to do with the Nobel prize? Well, even though everyone is driving optimally, there can still be apparently unexplainable traffic slowdowns on the highway. Even worse, there can be unexplainable slowdowns on your side of a divided highway even if the original problem is on the other side. How is this so? The answers are propagation and correlation.

Suppose there is an "event" on your side of the highway. The first driver to encounter it reacts optimally in some way, say by braking. But then there is the next driver - they don't even need to encounter the "event" to understand that they may need to brake too. The perpetuation of braking is an example of a propagation mechanism. One small "event" can be propagated into a traffic slowdown. And those slowdowns can develop a life of their own. The can persist long after the initial "event". They can be much bigger than the initial "event". And traffic engineers will even tell you that slowdowns can migrate forward or backward down the highway.

Further suppose that the "event" is on the other side of the highway. If you see someone slowing down going the opposite way on a divided highway, is it sometimes optimal to hit your brakes? You'd better believe it - for example there might be deer running across the highway (or if you are in NOLA a sleeping bear may have just fallen off the back of a circus truck). The upshot is that the "event" that causes a traffic slowdown doesn't even have to occur in the place that the slowdown does. This is correlation.

Kydland and Prescott's insight was that propagation and correlation can turn collections of individually optimal choices into bigger negative (and positive) events that can take on a life of their own.