Ah, the blogosphere is a wonderful thing. I asked for thoughts about the status of AS-AD in the intro macro textbook, and I got a whole bunch. I really appreciate the response! Here is my takeaway:
1. There is a lot of love out there for AS-AD. I didn’t want to get into a debate over its merits (and yes I know Romer’s version, the claim that AS is a kind of Phillips Curve, and so on), but I was curious about whether economists feel positively enough about the model to actually use it in real life (as opposed to the classroom). The answer is that some do. That said, I think one could follow policy debates in economics without any exposure to AS-AD: all the main arguments on every side—Keynesian, classical, Post Keynesian, Austrian—can be presented and normally are presented without recourse to the AS-AD format. There is a bit of AS-ADing on the web; there is little to none, as far as I know, in the corridors of policy.
2. AS-AD’s greatest claim to relevance may be its role in allowing the 1970s experience in the US to be placed in a different box than previous and subsequent episodes of macro distress. (This is Paul Krugman’s core point.) I can see the logic of this. The 1970s was a crucial decade for macrotheory, and if all recessions were like the OPEC-induced variety the real business cycle model would be the only way to go. By differentiating between “demand-side” and “supply-side” crunches, we can ringfence the 70s. AS-AD is the easiest way to do this.
3. And what is the alternative? Surely we don’t want to suggest that a model without prices (like the venerable Keynesian Cross) can fill the bill? Whatever the detailed merits of AS-AD, at least it allows you to talk about output and prices at the same time.
So have I budged? Maybe, somewhat.
1. I still think AS-AD is logically flawed. That’s a different debate, but I think the flaws show up in a widespread perception that the framework isn’t very useful for policy purposes. Here are two examples: (a) A recurring question is “what should the central bank do?” Assuming that the CB engages robotically in inflation targeting (the Romer version of the model) isn’t exactly the best starting point for trying to answer it. (b) The standard model of a shifting short run Phillips Curve (when unemployment is above or below NAIRU) provides a more empirically relevant basis for understanding the relationship between output and inflation than an AS curve. You can actually measure NAIRU, debate its stability, etc. That will give you an AS curve too. But what is the empirical basis for a shift in the AS curve? That’s about a shift in NAIRU, right? Am I the only one who notices that the literature that would justify and calibrate such shifts is thin and inconclusive? (I say this even though I think I have a rough understanding of the factors that might push NAIRU one way or the other.) A lot has been written on the impact of Obama’s initial stimulus on output and employment, for instance; how much has been written on its impact on NAIRU?
2. It’s true that without AS-AD it’s hard to put the 1970s in a separate box. How much that matters depends on how you want to respond to the new classical critique that emerged at the same time. One response is, OK you guys were right, but mostly about this one period, which really was a real business cycle. That’s not the only option.
3. What the alternative is depends on whether you think you always need a single model to tie everything together. That function is served by DSGE at the upper reaches of macrotheory. (Again, the merits of DSGE are not at issue at the moment.) But DSGE can’t do this job at the intro level. So do you tell a number of stories that bear on DSGE-ish issues, or do you opt for a simple model whose value is mainly to provide a thread that runs through the intro textbook? The mantra for textbooks today is simple stories with lots of repetition. Whether you can get introductory students to wrap their minds around multiple stories that stand in some tension with one another is a practical question. We’ll see.