There was one post yesterday, “Tl;dr and Modern Macroeconomics (Wonkish):”
(For the non-Twitterati “tl;dr” is short for “too long, didn’t read”.) Now here’s the Prof.:
Dear spell-correct: no, I do not mean “monkish.”
A short break from textbook revision, with macroeconomics and how to teach it still on my mind. So let me return to an old topic, the continuing usefulness of Hicksian IS-LM economics, in a somewhat different context.
When the Great Recession struck, there was a sharp division in economic commentary between those who had learned and appreciated the old Hicksian framework and those who hadn’t and/or didn’t. For that framework made some important predictions — namely, it said that things would be different at the zero lower bound. Increases in the monetary base — even huge increases — would not be inflationary. Budget deficits would not drive up interest rates. And fiscal multipliers would be much larger than they are in normal times, when fiscal expansion or contraction is offset by monetary policy.
Those were deeply controversial predictions at the time, but they were overwhelmingly vindicated by experience — dead-enders are reduced to arguing that Hicksians just happened to be right for the wrong reasons.
But here’s the thing: doing anything like HIcksian analysis in public is still very much frowned on within the economics profession. It’s ad hoc, not microfounded, sloppy about intertemporal relationships. DSGE models with sticky prices are OK; publishing IS-LMish stuff, even in a policy forum, remains hard and in general is possible only for old guys with enough professional capital to get away with it.
So how much is lost as a result? What set me off was reading Eggertsson et al (EMSS) on contagious secular stagnation. It’s serious work, and I agree with the main conclusions; I am also a big admirer of all of the economists involved, particularly Gauti, who was investigating the weird economics of the liquidity trap long before it was cool.
And yet … it’s really tough going, epitomizing too long; didn’t read to the nth degree. In part I guess I’m just an aging economist with much less tolerance for algebraic grinding than I used to have. You also want to bear in mind the old principle that the optimal level of technical difficulty in papers is always precisely the level of your own papers. But still.
What do we get out of the rigor — the overlapping-generations setup, the explicit modeling of borrowing constraints, and so on? Suppose you came at this issue in an old-fashioned way, using Mundell-Fleming — the open-economy version of IS-LM. You would boil it down (as Olivier Blanchard has suggested in an email) to a Metzler diagram, with the exchange rate (price of foreign currency) on the horizontal axis and interest rates on the vertical:
The idea here is that a depreciation of Home’s currency causes economic expansion in Home, which Home’s central bank leans against, hence the upward slope; meanwhile, it causes contraction in Foreign, which Foreign’s central bank also leans against, hence downward slope. But both face a potential zero lower bound, hence the flat sections.
With perfect capital mobility and static expectations, interest rates must be equalized, so equilibrium is where the two lines cross. And it’s now obvious that an adverse shock in Foreign, suggested by the blue arrows, will push interest rates down in both countries. If the shock is enough to drive Foreign to the ZLB, it will do the same to Home, as transmitted through the exchange rate. In other words, Europe can export its secular stagnation to us via a weak euro and a strong dollar.
Now, you get a few additional insights from the EMSS paper, such as the role of credit constraints in inducing stagnation and the rule of limits on capital mobility in limiting its spread. But the cost in terms of complexity and cumbersomeness is huge.
And this cumbersomeness may even lead to loss of insight. The paper relies, necessarily, on the analysis of steady states. Yet I would argue that the transmission of the liquidity trap depends crucially on how permanent the shock is perceived to be — which is an insight you lose by assuming a steady state.
But, some readers may say, haven’t I myself used this kind of framework, both in my original liquidity trap analysis and in work with Gauti on deleveraging? Yes indeed — and while part of the reason was to get through the anti-Hicks barrier, in each case I believed that dotting those i’s and crossing those t’s yielded some valuable insights. In fact, I didn’t believe in the liquidity trap until I saw it pop up in a New Keynesian model, and doing the deleveraging math really helped clarify my thinking there too.
So I don’t have any general opposition to the more elaborate modeling approach. What worries me is the effective prohibition on simple, ad hoc models that sometimes yield most of the insight — in the case of contagious secular stagnation, I’d put the ratio well above 90 percent — in a form that is much more useful for real-world policy discussion.
Or then again, maybe it’s just my vintage. Also, you kids get off my lawn.