You can follow this conversation by subscribing to the comment feed for this post.

This may not be your intent but I associate the red line in both displays as being the AD line for boom examples in the early stage of the boom.

I associate the green line with monetarily stable (constant money supply) supply-demand relationships where more supply results in lower prices AND increased consumption (more potatoes results in a lower potato price so more people eat potatoes).

Both diagrams seem logical.

The next challenge is to decide where the AS and P=E(P) should CORRECTLY be placed. To explain, we built our model (both diagrams) with the expectation of general conditions; the red and green lines are general expectation of departure from stability assuming that the starting point is stable. Now it is very UNLIKELY that the present economic point is in a stable position, so, we must decide if the vertical AS line (the supply line) is (at present) following red curve conditions or green curve conditions. In the same fashion, we must decide if the P=E(P) line (at present) is following the red or green expectation model.

We can apply this model divergence to the residential construction period of 2007-8. Coming into the 2007-8 period, bankers and government seemed to be following the red line models. Then, suddenly, both bankers and government seemed to decide that they should be following the green line model. At that point in time, a considerable difference had accumulated between the expectations of the red and green models. There was a sudden adjustment to the whole economy as the dynamics of finance tried to adjust to the more accurate (in the long term) green line model. That adjustment seems to be continuing as we write.

I like the line of logic presented in this post. I wonder if can endure and expand?

Nick Rowe: "Equilibrium" does not (necessarily) mean "quantity demanded equals quantity supplied". It simply means "that which the model predicts". (They only refer to the same thing in a model that assumes continuous market-clearning)."

If nothing is equal in an equilibrium, then better to find a new term. ;) But let's grant the new definition. :)

Nick Rowe: "And that's central to the critique of the old view. If your model says that X* will happen, how can you talk about what would happen if X were different from X*? According to your model, X can't be different from X*. You are contradicting your own model."

Excuse me. The are such things called time and error. Anyone who imagines that an economic model fit the data perfectly has a screw loose. And if the data deviate from the prediction of the model at any point in time, then the model should predict that the data should approach the prediction as time goes by, or it is not a prediction.

Nick Rowe: "When old economists talk about "stability", we are talking about departures from the model's equilibrium. When young economists talk about "stability", they are talking about movements of the model's equilibrium. Those are very different things."

Indeed they are. Another word for "departure" is "error", the difference between prediction and data. Another word is "perturbation", errors which have causes that are not part of the model. It sounds like old economists paid more attention to concepts like error and perturbation than new economists. But without the concept of error, there is no prediction, there is merely assumption. So perhaps the new concept of equilibrium is not what the model predicts, but what it assumes.

Min: "If nothing is equal in an equilibrium, then better to find a new term. ;) But let's grant the new definition. :)"

Things usually are equal in a model's equilibrium. But those things aren't always quantities demanded and quantities supplied. Plus, it's too late to change how we use the word.

Nick Rowe: "Plus, it's too late to change how we use the word."

Not that it is any of my business, but you have just said that the usage of the word has changed. Why should it not keep changing? (Or has its usage reached an, ahem, equilibrium? ;))

Let me add a point about stability, perturbation, and feedback. A system in which perturbations produce negative feedback is stable. Deviations lessen over time. A system in which perturbations produce positive feedback is unstable. Deviations increase over time. We have known since the 1990s that the solar system is unstable. But that is only in the long run. In the short run of tens of thousands of years, it is stable. Just because we can used the theory of gravity to predict the trajectories of planets in the solar system does not mean that it is stable.

A few more mathie points, I suppose:

*) A system is only stable or unstable if it has dynamics. Accounting identities describe what things are, not how things change. (MV = PY is an accounting identity; saying "increase the money base to increase the nominal output" is a dynamical story that makes assumptions about V)
*) Unstable does not mean unpredictable, as Min points out. But instability does limit "really long run" predictability. Theories which don't constrain the price level are unstable over time even if they are predictable, since they're consistent with P=1 or P=1 billion at infinity.
*) Models of a time-evolving equilibrium are useful if and only if that equilibrium is stable to perturbations
*) Models with stable equilibriums can be simpler, since they only need to show that the dynamics return to equilibrium and are less sensitive on how (or how quickly) things get back to "normal."
*) Instability is harder, since you need a very short-term and detailed model of how things go awry.
*) The existence of business cycles suggests that the economy is weakly unstable. History suggests that the ultimate constraints are political, in that people will reject an economy that stops working too badly.
*) Proper control can turn an unstable equilibrium into a stable one, just like balancing a yardstick in the palm of your hand. Monetary policy is an attempt to do this with the economy.

Majromax: "Proper control can turn an unstable equilibrium into a stable one, just like balancing a yardstick in the palm of your hand. Monetary policy is an attempt to do this with the economy."

I would rephrase that. We cannot say that the economy is unstable, but monetary policy tries to make it stable. That makes no sense. Because there is always a monetary policy of one kind or another. What we should instead say is: whether the economy is stable or unstable depends on the monetary policy being followed.

Now, stop all this engineering/math talk for a minute, and look at my top picture. What do you see? What do you see as the difference between the economy with the red AD curve and the economy with the green AD curve?

Min: "Another word for "departure" is "error", the difference between prediction and data. Another word is "perturbation", errors which have causes that are not part of the model. It sounds like old economists paid more attention to concepts like error and perturbation than new economists."

No. That is missing the point.

Look at the top diagram. There are two sorts of errors we can make:

1. We can make errors in understanding the things that shift the AD and AS curves, so they shift when we don't expect them to, or don't understand why they shift. Old and new economists are exactly the same with those sorts of errors. We know we will make them. That's why we put an error term in all our equations.

2. We can make an error in saying that the economy is always at the intersection of the AD and AS curves.

The second sort of error has very different effects, depending on whether we have the green or red AD curve.

> I would rephrase that. We cannot say that the economy is unstable, but monetary policy tries to make it stable. That makes no sense. Because there is always a monetary policy of one kind or another.

Good point, my language was sloppy. The economy, under current and historical monetary policies (including precious metal standards) is weakly unstable. It is the goal of current monetary policy to actively make the economy more stable(*).

"There is always a monetary policy" also speaks to the difference between the AD curve itself and a point along that curve. We obviously can't observe the entire curve, just our current (or historical) place along it. Since the curve itself is useful, it's probably sensible to say "monetary policy does nothing when aggregate demand moves along the preconceived curve; monetary policy does something when the curve shifts."

> Now, stop all this engineering/math talk for a minute, and look at my top picture. What do you see? What do you see as the difference between the economy with the red AD curve and the economy with the green AD curve?

The red curve describes a world where the gross quantity of goods demanded goes up with the price level. The green curve describes a world where the gross quantity of goods demanded goes down with increases in the price level. The blue curve describes a world where the gross supply of goods is independent of both the price level and aggregate demand(**).

By themselves, those lines are just curves on a graph, since there's no dynamical story.

But if you combine this with market prices, where in aggregate the economy behaves consistently with micro-level "prices of goods increase when there is a shortage," then we immediately get your conclusion that the equilibrium of (red,blue) is unstable whereas (green,blue) is stable.

Other price-setting stories may give different results; for example a price-control regime would find that *any* price is stable, but it would just lead to persistent shortage/surplus if the price was not at the intersection of the curves.

(Mathie speak: the system isn't closed without a theory of the price level/inflation rate. From just the graph, you have AS and AD as functions of P, but no constraint on how P changes as a function of the output gap. This is where the classical Philips Curve comes in.)

(*) -- That's worth mentioning specifically because that hasn't always been the case. Breaking with the idea that the central bank exists to finance the government is a pretty big structural change.

(**) - Almost always a structural given, but worth mentioning anyway.

Moi: "Another word for "departure" is "error", the difference between prediction and data. Another word is "perturbation", errors which have causes that are not part of the model. It sounds like old economists paid more attention to concepts like error and perturbation than new economists."

Nick Rowe: "No. That is missing the point."

Nick Rowe: "And that's central to the critique of the old view. If your model says that X* will happen, how can you talk about what would happen if X were different from X*? According to your model, X can't be different from X*. You are contradicting your own model."

To say that X can't be different from X* is to ignore errors. And that makes it a bullshit critique. :)

Nick Rowe: "What do you see as the difference between the economy with the red AD curve and the economy with the green AD curve?"

An economy with the green AD curve is one where deviations are dampened by negative feedback. An economy with the red AD curve is one which is inherently unstable, because deviations are magnified by runaway feedback. And deviations always occur.

Min: OK, you get it. You just talk funny. Probably engineer-speak!

Just because I have defended the "old" idea of stability does not mean that I deny the new one. In fact, my example of the solar system is one of new instability. We assume that we can predict the motions of the planets accurately far into the future, and that our model is correct. Then we find instability that is not the result of deviation from the model. BTW, in the case of the solar system, chaotic is a better term than unstable.

Nick Rowe: "Min: OK, you get it. You just talk funny. Probably engineer-speak!"

I do talk funny, now that you mention it. ;)

Systems speak, I would say. :) Actually, I am more a student of human systems than other kinds.

The comments to this entry are closed.

• WWW