Lately, I have been reading a paper by Roger Backhouse and David Laidler, “What Was Lost with IS-LM” (an earlier version is available here) which was part of a very interesting symposium of 11 papers on the IS-LM model published as a supplement to the 2004 volume of History of Political Economy. The main thesis of the paper is that the IS-LM model, like the General Theory of which it is a partial and imperfect distillation, aborted a number of promising developments in the rapidly developing, but still nascent, field of macroeconomics in the 1920 and 1930s, developments that just might, had they not been elbowed aside by the IS-LM model, have evolved into a more useful and relevant theory of macroeconomic fluctuations and policy than we now possess. Even though I have occasionally sparred with Scott Sumner about IS-LM – with me pushing back a bit at Scott’s attacks on IS-LM — I have a lot of sympathy for the Backhouse-Laidler thesis.
The Backhouse-Laidler paper is too long to summarize, but I will just note that there are four types of loss that they attribute to IS-LM, which are all, more or less, derivative of the static equilibrium character of Keynes’s analytic method in both the General Theory and the IS-LM construction.
1 The loss of dynamic analysis. IS-LM is a single-period model.
2 The loss of intertemporal choice and expectations. Intertemporal choice and expectations are excluded a priori in a single-period model.
3 The loss of policy regimes. In a single-period model, policy is a one-time affair. The problem of setting up a regime that leads to optimal results over time doesn’t arise.
4 The loss of intertemporal coordination failures. Another concept that is irrelevant in a one-period model.
There was one particular passage that I found especially impressive. Commenting on the lack of any systematic dynamic analysis in the GT, Backhouse and Laidler observe,
[A]lthough [Keynes] made many remarks that could be (and in some cases were later) turned into dynamic models, the emphasis of the General Theory was nevertheless on unemployment as an equilibrium phenomenon.
Dynamic accounts of how money wages might affect employment were only a little more integrated into Keynes’s formal analysis than they were later into IS-LM. Far more significant for the development in Keynes’s thought is how Keynes himself systematically neglected dynamic factors that had been discussed in previous explanations of unemployment. This was a feature of the General Theory remarked on by Bertil Ohlin (1937, 235-36):
Keynes’s theoretical system . . . is equally “old-fashioned” in the second respect which characterizes recent economic theory – namely, the attempt to break away from an explanation of economic events by means of orthodox equilibrium constructions. No other analysis of trade fluctuations in recent years – with the possible exception of the Mises-Hayek school – follows such conservative lines in this respect. In fact, Keynes is much more of an “equilibrium theorist” than such economists as Cassel and, I think, Marshall.
Backhouse and Laidler go on to cite the Stockholm School (of which Ohlin was a leading figure) as an example of explicitly dynamic analysis.
As Bjorn Hansson (1982) has shown, this group developed an explicit method, using the idea of a succession of “unit periods,” in which each period began with agents having plans based on newly formed expectations about the outcome of executing them, and ended with the economy in some new situation that was the outcome of executing them, and ended with the economy in some new situation that was the outcome of market processes set in motion by the incompatibility of those plans, and in which expectations had been reformulated, too, in the light of experience. They applied this method to the construction of a wide variety of what they called “model sequences,” many of which involved downward spirals in economic activity at whose very heart lay rising unemployment. This is not the place to discuss the vexed question of the extent to which some of this work anticipated the Keynesian multiplier process, but it should be noted that, in IS-LM, it is the limit to which such processes move, rather than the time path they follow to get there, that is emphasized.
The Stockholm method seems to me exactly the right way to explain business-cycle downturns. In normal times, there is a rough – certainly not perfect, but good enough — correspondence of expectations among agents. That correspondence of expectations implies that the individual plans contingent on those expectations will be more or less compatible with one another. Surprises happen; here and there people are disappointed and regret past decisions, but, on the whole, they are able to adjust as needed to muddle through. There is usually enough flexibility in a system to allow most people to adjust their plans in response to unforeseen circumstances, so that the disappointment of some expectations doesn’t become contagious, causing a systemic crisis.
But when there is some sort of major shock – and it can only be a shock if it is unforeseen – the system may not be able to adjust. Instead, the disappointment of expectations becomes contagious. If my customers aren’t able to sell their products, I may not be able to sell mine. Expectations are like networks. If there is a breakdown at some point in the network, the whole network may collapse or malfunction. Because expectations and plans fit together in interlocking networks, it is possible that even a disturbance at one point in the network can cascade over an increasingly wide group of agents, leading to something like a system-wide breakdown, a financial crisis or a depression.
But the “problem” with the Stockholm method was that it was open-ended. It could offer only “a wide variety” of “model sequences,” without specifying a determinate solution. It was just this gap in the Stockholm approach that Keynes was able to fill. He provided a determinate equilibrium, “the limit to which the Stockholm model sequences would move, rather than the time path they follow to get there.” A messy, but insightful, approach to explaining the phenomenon of downward spirals in economic activity coupled with rising unemployment was cast aside in favor of the neater, simpler approach of Keynes. No wonder Ohlin sounds annoyed in his comment, quoted by Backhouse and Laidler, about Keynes. Tractability trumped insight.
Unfortunately, that is still the case today. Open-ended models of the sort that the Stockholm School tried to develop still cannot compete with the RBC and DSGE models that have displaced IS-LM and now dominate modern macroeconomics. The basic idea that modern economies form networks, and that networks have properties that are not reducible to just the nodes forming them has yet to penetrate the trained intuition of modern macroeconomists. Otherwise, how would it have been possible to imagine that a macroeconomic model could consist of a single representative agent? And just because modern macroeconomists have expanded their models to include more than a single representative agent doesn’t mean that the intellectual gap evidenced by the introduction of representative-agent models into macroeconomic discourse has been closed.
Uneasy Money 
David,
I like the Stockholm picture where incompatible expectations cause economies to go into recession (expectations either have a neutral impact on e.g. NGDP, or a negative impact). This is reminiscent of Friedman’s plucking model.
Yet another way to see incompatible expectations is that the price mechanism fails to move information from the demand to the supply — this also produces the effect that there is either a neutral or negative impact. Information received by the supply can only be less than or equal to the information sent by the demand: I(S) ≤ I(D)
http://informationtransfereconomics.blogspot.com/2014/05/the-effect-of-expectations-in-economics.html
In this picture, expectations are generally met (or people muddle through) so that I(S) ~ I(D), but in cases of recessions, I(S) < I(D) … the information in the plans isn't getting through the market (e.g. because they are mutually inconsistent, so that information is lost in some, maybe all, plans).
ps You can also get supply and demand diagrams from assuming I(S) = I(D) and that either I(S) or I(D) is constant.
LikeLike
Hey Prof. Glasner,
I’m Suvy and I’ve been following your blog for a decent while. I’ve also read your book Free Banking and Monetary Reform and I found it very interesting to say the least. Anyways, I did a post on IS/LM recently which actually culminated a series I did on money and banking. I start off with the basics (loans creating deposits), then go on to speak about different monetary policy regimes, then did a post on the impact of QE, and finished it off with a post on IS/LM. I’ve attached the posts in order as links below.
http://suvysthoughts.blogspot.com/2014/07/money-and-banking-basics.html
http://suvysthoughts.blogspot.com/2014/07/monetary-policy-basics-and-qe.html
http://suvysthoughts.blogspot.com/2014/08/impact-of-qe.html
http://suvysthoughts.blogspot.com/2014/08/why-islm-sucks.html
Anyways, IS/LM is a very poor model. Although it gets some things right, it’s the exact wrong way to look at, and think about, monetary policy. I’d like to get your take on my view of things. Your blog is really cool BTW.
LikeLike
David, what do you think of models like this (http://www.dklevine.com/archive/refs4786969000000000359.pdf), this (http://core.kmi.open.ac.uk/download/pdf/6715710.pdf) or this (https://repositori.upf.edu/bitstream/handle/10230/20777/1208.pdf?sequence=1) as networked based macro paradigms? I’d be interested to hear your thoughts
LikeLike
David,
You are sounding more and more like a complexity theorist! There is a science of how agents self-organize and how agents behave. As you point out, though, such disciplines can never produce the tidy equilibrium solutions of economics. DGSE, as a lamppost, gives off a reassuring light. Unfortunately, the keys often lay elsewhere.
LikeLike
Meant to say “…how agents self-organize and how networks behave” above
LikeLike
Jason, Thanks for the link and the explanation. How do you know which direction information is moving?
Suvy, Thanks for the links, which I have only had a chance to skim through quickly. I agree that IS-LM is not a very good model, though it can be used to get some important points across. But you need other models to know how to use IS-LM safely. I disagree that IS is derived from the identity of savings and investment. That identity, like all accounting identities, tells us nothing about the real world. I = S is an equilibrium condition which may or may not be true. Keynes, unfortunately, had a blind spot about the difference between identities and equilibrium conditions which caused him to get badly confused in a number of places. Hawtrey kept trying to set him straight, but on that particular point at least Keynes was a bad student.
Jonathan, Thanks for the links. I am aware of the paper by Acemoglu et al. but haven’t read it. But I am glad to see that a very high-powered theorist like him is thinking seriously about this problem, so I am hoping that the idea will catch on. My only reservation is that they seem to think that the macro shock must originate in a micro-productivity shock. I think it’s also possible for a macro disturbance to be transmitted and amplified by sectoral linkages.
Diego, Can you suggest a good primer on complexity theory?
LikeLike
David,
Melanie Mitchell’s book. She is a computer scientist and researcher at the Santa Fe Institute.
LikeLike
I was an econ PhD student at Yale in the 1980s, so had the benefit of seeing Tobin manipulate IS-LM, which was masterful. He had the model in his bones. He explained it not as a 1-period model but as a snap-shot of the economy, For a proper analysis of the economy, you would use it much like this sequence of pictures: http://en.wikipedia.org/wiki/Sallie_Gardner_at_a_Gallop , or rather, he would, indicating how the lines moved from period to period. Alternatively (IIRC), it is a model of short period equilibrium, during which prices adjust but stocks (inventories, capital, whatever you like) do not.
LikeLike
marcel, It is always a pleasure to watch a master at work, so you were lucky to have seen Tobin in action. I understand that you can think of IS-LM as a series of snapshots, but it seems to me that it would be preferable to have an explicitly dynamic model that incorporates expectations. Of course it is possible to incorporate expectations into the IS-LM framework, but, as you point out, it doesn’t allow for an explicit analysis of real capital. I have a series of posts on Earl Thompson’s reformulation of macro theory in which he substituted a market for real capital services in place of the Keynesian expenditure function or investment equal savings equilibrium condition (IS curve). You might want to have a look at those. At some point, I may try to continue that series of posts.
LikeLike
Hi David,
Sorry I’m late to responding to your question (I think I forgot to check the notify me of comments box). You asked:
“How do you know which direction information is moving?”
One potential explanation is that demand is relatively hard to measure (without a price mechanism) compared to supply (in the simplest form of an economy, counting up the the goods). Therefore demand information seems vulnerable to being inefficiently collected by the market i.e. I(S) < I(D).
In the case where I(D) = I(S), the direction doesn't really matter (it comes down to a sign convention).
This explanation partially derives from discussion with Peter Fielitz (one of the originators of the information transfer framework as it applies to physics — credit where credit is due). It's still a work in progress — especially in the case where I(S) < I(D); I'd love to get criticism from a few economics Phds not just on this but some of the results that come out the framework such as this:
http://informationtransfereconomics.blogspot.com/2014/03/the-islm-model-again.html
Cheers,
Jason
LikeLike
Thank you David.Very interesting !
I assume that Stockholm-method
refer to Erik Lundberg, and his use of sequence analyzes to study
the business cycle as it is put forward in his 1937 dissertation
Theory of Economic Expansion?
Erik Lundberg is sadly almost forgotten today
although he defintly was one of the great economists
of the 1900 century with an impact on both theory and policy.
One example : Paul Samuelson´s famous Oscillator and multiplier-accelerator model is basically
just a mathematical formalisation of what Erik Lundberg
allready have expressed in Theory of Economic Expansion.
Erik Lundberg dissertation is still worth reading in my view
but even more i think his
“The Development of Swedish
and Keynesian Macroeconomic Theory and
its Impact on Economic Policy-Erik Lundberg ”
Cambridge University Press, Cambridge 1996
and ”
Studies in Economic Instability and Change-Erik Lundberg ”
SNS förlag, Stockholm 1995″ to just name two.
LikeLike
Excellent October 2013 analysis by Hummel that I can’t remember any Market Monetarists commenting on…….
“The Myth of Federal Reserve Control Over Interest Rates”
http://www.econlib.org/library/Columns/y2013/Hummelinterestrates.html
LikeLike