Update: Paul Krugman offers comment here. Similar sentiments.
Mark Thoma comments on Peter Dorman’s reaction to the paucity of AD-AS models in the blogosphere. I think I can offer a unique perspective because (a) I’ve only completed a high-school level econ class, (b) I form a lot of my opinions based on discussion in the blogosphere, and (c) I blog a fair amount myself.
I have a pretty solid mathematical intuition, but lack the experience (or perhaps smarts, who knows) to understand DSGEs at a granular level. I can get through most of this, but probably can’t teach it to anyone which has to be the real indicator of knowledge. Dorman implies that if not for this restriction, the DSGE would be a better tool.
But I am sufficiently comfortable with my statistics to know DSGE’s suck (“suck” being a very technical term whose math I will leave as an exercise to the reader) on out-of-sample forecasts. That’s the first beauty of AD-AS, it’s not trying to “forecast” anything, but rather abstract thematic elements of the economy.
Here’s what it can do (read this as an uneducated high schooler’s defense of a too-complicated subject):
- Differentiate between supply and demand shocks. Anyone who knows the model can understand the difference between unemployment from an oil embargo and financial crisis.
- Convey long-run neutrality of money, which isn’t as natural without the correcting force within AD-AS.
- And with a bit of salt and pepper, we can approximate the real economy from the rather abstract “neutrality” point of view with hysteresis as the transmission between AD and LRAS.
- Evade bullshit. It seems you can create a DSGE for anything under the sun. But for now I feel relatively confident AS affects prices and AD affects quantity: that is, AD is not perfectly elastic, removing relevance in crisis thereof. (Update: and, as Nick Rowe shows, it’s very flexible too).
But here we reach it’s fundamental limitation:
- Who really thinks AD-AS (or for that matter IS-LM) are independent of each other? It’s absolutely unintuitive to think so, and that’s a dinger for a so-called “intuitive” model.
- AD/AS and IS-LM may be prone to stock-flow considerations. Intertemporal optimization is key here, and actually that’s a lot easier to grasp (mathematically, but definitely intuitively) as an introductory student. (Irving Fisher’s math is rather more enjoyable than modern econ papers, Milton Friedman’s too).
IS-LM isn’t the same as AD-AS, but if I had to bet Dorman (or most in the anti AD-AS camp) would place the older version right with AD-AS relative to microfounded modern models. True, it tries to add dynamism to a staticky AD-AS and without intertemporal considerations it’s prone to the same criticisms.
But it gives a rock solid intuition of the real economy. LM (and AD too) is actually really confusing in the context of central banks, but once you add wage rigidity you have a pretty close approximation to old Keynesian thinking. If you think Keynes was an important economist, this is an end in and of itself.
But, as the archetypical two-handed economist, there’s a flip:
- AD-AS (at least at an AP level) is poorly taught. Stock-flow considerations are completely ignored.
- The role of monetary and fiscal policy as “demand-side” are obscured. (IS-LM with rigidity is a quick fix to this.)
Ultimately, AD-AS anchored my thought regarding the economy, for better or worse. I understand the caveats, and having blogged for a few months I think I have a more nuanced understanding thereof. Two lines can’t forecast anything, life would be boring if they could. But they are not a “placeholder” for something better to come.
In terms of informing policy, the dominant positions (old Keynesian and market monetarist) are well-explained within an AD-AS or IS-LM framework. Microfounding old thinking through monopolistic New Keynesianism with menu costs and all seem to have enriched economic thought. The question is, can it better drive policy? More importantly, pedagogically, can it import a better understanding thereof?
Oh, and reading through the blogosphere, it doesn’t seem like people explicitly state AD-AS/IS-LM as their mode of thinking, but that’s almost all you need to understand most concepts, at least thematically. For me, that’s enough to conclude that AD-AS isn’t just an artifact, but a great heuristic employed by the best thinkers. They understand its caveats and use it with caution. But its back-of-the-envelope attraction is what makes it so valuable.