Should we teach IS/LM?

With this year’s Nobel having gone macro, Offsetting needs a macro post.

I have been enjoying the discussion about the IS/LM model in the blogsphere. It started with Tyler Cowen here discussing why he doesn’t like IS/LM. Brad Delong, Krugman, Scott Sumner and others responded in different ways, pro and con (see the links at Steve Landsburg's take on the issue here). WCI’s Nick Rowe makes the point that in many ways, the argument is one of semantics—“It's still IS/LM to me”.

For me, the interesting question is not whether IS/LM is a useful abstraction for some questions, but how (and when) it should be taught. After all, academic economists have access to the full range of models, but typical economics graduates will be highly constrained by the basic models with which they were introduced to macroeconomics. So, I am going to imagine that I were able to write some rules to guide the content of undergraduate macroeconomics texts and ask what role IS/LM would play in those rules. This is what I came up with.

1. Start with the determinants of long-run economic growth.

Depending on the level, this could be a long-run dynamic general equilibrium model or just a simple presentation of growth accounting. The key thing is to emphasise the role of investment and, by extension, saving rather than fiscal and monetary policy in determining long-run growth. This was important during the great moderation, but it is arguably even more important now when we want to emphasise that the current situation is not normal and so policy prescriptions need to start from an analysis of what makes the situation different.

2. Spend a lot of time on the 45-degree line model when introducing the notion of aggregate demand.

When I first learnt macroeconomics, it was standard to start with the Keynesian cross in both principles and intermediate courses, but it seems to no longer be in fashion. I think that is a mistake. Both the IS and the AD curves show combinations of variables that have equilibrium properties, but both look like they should be read as a causal behavioural relationship from the horizontal to vertical axes, just like a micro demand curve. The 45-degree line model contains a causal, behavioural relationship and an important accounting identity. Students need to understand the difference between them thoroughly, how the economy will always be on the 45-degree line, and how undesired changes in inventories (or potentially quantity rationing) explain how the economy can exist away from the equilibrium point. The diagram is also a useful way of conveying the notion of hot-potato money.

3. Go straight from the 45-degree line to the AD/potential-output diagram.

Cut to the chase: this is the diagram that combines the LR determinants of output from step 1 (the vertical potential output line) with the Keynesian SR model from Step 2. It doesn’t matter what story you tell for why the equilibrium relationship between output and price is negative or how policy can affect the position of the AD curve. With this diagram, coupled with an expectations-augmented Phillips curve, one can discuss price dynamics, expectations, optimal Taylor rule type polices explaining why it is better to respond harder to aggregate demand shocks than to price shocks, etc.

Once this framework is in place, then, and only then, in my view, is it appropriate to explore graphical models of the determinants of aggregate demand, such as IS/LM. But before getting to that…

3a. …Remove any mention of the words “demand” and “supply” from this diagram…

Potential output is not “supply” in an analogous sense to a supply curve, indicating the willingness of producers to bring goods to market as a function of the price. Even more, the downward-sloping aggregate demand curve looks like a micro-style demand curve showing the amount buyers would like to buy at a given price, ceteris paribus, when in fact it is a locus of equilibrium points at which planned output equals actual output. Calling these curves aggregate demand and aggregate supply is guaranteed to confuse 90% of undergraduate students.

3b. ... And even with a different name, there is no place for the curve sometimes called short-run aggregate supply…

SRAS serves no useful purpose that I can see. One can describe price dynamics in terms of an economy moving along the AD curve without having a SR supply curve. I am open to being corrected here but as far as I can make out, the so-called SR aggregate supply curve is nothing more than the following: i) In the short-run, the economy will be on the so-called AD curve, but not necessarily on the vertical LRAS curve; ii) students, however, will want the location of the economy to be at the intersection of two curves, so we need a new curve; and iii) since the first curve has been misnamed aggregate demand and is downward sloping, we might as well label the other curve aggregate supply and make it upward-sloping.

3c….And why assume the economy is on the AD curve in the short-run?

The economy is only on the AD curve if it is at equilibrium in the 45-degree line diagram. Why assume that in the short-run there are no unplanned changes in inventories?

4. Return to your favourite version (or versions) of IS/LM.

Now that the basic ideas of macro are in place—long-run output depends on long-run capacity determination, short-run capacity utilisation may depend on the determinants of nominal spending, and the big policy questions concern how much one wants to manipulate nominal spending and how one brings that about—then is the time to get into details about liquidity traps, what interest rates the central bank can control, real versus nominal interest, the role of financial intermediation in the whole process, the possibly complicated feedback loops between SR nominal spending and LR capacity accumulation; etc. IS/LM is probably a useful diagrammatic tool to have in the mix here, but let’s do the basics first.