Steve Keen’s AS-AD Curves and a Suggestion For a New Stock-Flow Equilibrium Approach


A commenter on Lord Keynes’ blog recently called my attention to something rather interesting; namely, that Steve Keen seems to be using some sort of supply and demand framework to determine price in the macroeconomy in his models. Let me just say that I do not follow Steve’s work all that closely and so I apologise if this is old news and has since been overcome. With that caveat, a few comments.

The moment I heard this I thought, “Ah, Steve must be using the old aggregate supply/aggregate demand (AS-AD) framework…”; indeed, I responded as such to the commenter. He then directed attention to the following article in which Keen explains that when he was integrating prices into his debt model he found himself with a number of paths that he could take.

He could take the neoclassical path and equate marginal revenue with marginal cost. But that would, as Steve well knows, be cheating. Alternatively, he could adopt a Post-Keynesian perspective and use a mark-up pricing framework. He says that this would have been a ‘fudge’ but does not say why. I would imagine that he thought that this would be a fudge because it would have been very difficult to get substantial price movements in, for example, financial markets if there was a crash.

He instead opted for a third path which he describes as such:

The one way I could do that was to argue that the price level would adjust under the pressure from the flow of monetary demand on one side, and the pressure of physical supply on the other.

This seems to me to be identical to the old AS-AD framework which can be seen in the diagram below.

AS-ADcurvejpgWhat the AS-AD framework shows is a trade-off between prices and output. The idea is that as aggregate demand increases both output and price will rise. As we can see the aggregate supply curve is quite flat at low levels of output while it is basically vertical at high levels. This indicates that at low levels of output an increase in aggregate demand will lead to large increases in output with very small increases in prices because the economy is assumed to have significant excess capacity, while at high levels of output an increase in aggregate demand will only affect prices as the economy is assumed to be at full capacity.

The AS-AD is not the worst framework imaginable. Indeed, some Post-Keynesians, like Paul Davidson, have advocated its use for didactic purposes. But there are many problems with it. For one, the downward-sloping aggregate demand curve that you see in the above graph is derived from an ISLM curve that assumes a linear relationship between the interest rate and output and a fixed supply of money by the central bank.

It is assumed that as prices rise the real interest rate increases because the real value of money falls and so higher prices lead to a fall in real output via the rise in the interest rate; this is what the downward-sloping AD curve depicts. Of course, this rests on the crucial assumption that the central bank sets a fixed supply of money. Yes, we can shift the supply of money in the model by shifting the money supply curve which will then shift the AD curve, but we cannot avoid the simple fact that the AS-AD model is not compatible with Post-Keynesian endogenous money theory.

Basically, any objections to the ISLM framework that Post-Keynesians might hold equally apply to the AS-AD framework. There is simply no getting away from this. If we want to imagine the price level in the macroeconomy as simply being based on the interaction of a giant aggregate supply and aggregate demand curve we must accept the ISLM. My sense is that Keen will not be very happy about doing this.

I would also say that the AS-AD framework does not properly incorporate expectations. Surely this is an enormous problem for anyone trying to model speculative dynamics as I assume that Keen is. So, what is the alternative? Well, as I’ve said before on this blog, I tried to create a new theory of pricing for my dissertation that avoids these problems. Perhaps it would be better suited to such models.

Here, then, is a pricing equation for the macroeconomy in line with the theory laid out in my dissertation. As will shortly be seen it provides us with many advantages which I shall run through shortly. (I have omitted some more complex properties of the final equation as they would take up too much time to discuss here and will be discussed in the full working paper that is hopefully to be published soon. I should note, however, that these properties lead to some very interesting conclusions when seen in the context of the work of Hyman Minsky).

coverThose terms may seem a little obscure to most people so here is a table laying them out,

coverAs can be seen this is a framework that is not reliant on any notion of market equilibrium or supply and demand curves. Rather it is more akin to a Keynesian multiplier relationship. Or, to put that another way, it views price as the outcome of a stock-flow equilibrium process. We simply ‘plug in’ certain variables and we get price outcomes.

In the full framework the components of price are also broken down into various sectors — for example, financial asset prices are determined by the both the government sector (central banks etc. — think QE) and the private sector (financial market investors etc.) — which gives the framework even more flexibility.

The aggregate price level, which incorporates both asset prices and real prices which I distinguish in the paper as being those that contribute and those that do not contribute to Gross Domestic Product (yes, the framework is consistent with the national accounts) is determined both by the quantity of assets, real and financial, supplied (qZ) and the amount demanded. The latter is then broken down into the quantity of financial assets demanded based on future expected prices (Pef), the quantity of real assets demanded based on future expected price increases (Per) — these are the speculative components of demand — and finally, the quantity of assets demanded for ‘real’ consumption and investment, (Pr).

The framework incorporates the best parts of the supply and demand framework given allowances for quantity rather than price adjustments. This means that it can incorporate Post-Keynesian administered pricing theory without totally ditching supply and demand. It also incorporates Marshallian price elasticity concerns and expectations as they exist in both Post-Keynesian and behavioral economics. Taken together I believe that we can reduce price formation purely to these variables and to no more. I also believe that this framework can be applied to any price formation that takes place in any type of economy.

In the paper in which I present the theory I refer to it as a ‘general theory of pricing’ and in that regard I have the following quotation from Keynes in mind which he wrote in the forward to the German edition of the General Theory (which I discuss in further detail here),

This is one of the reasons that justifies the fact that I call my theory a general theory. Since it is based on fewer hypotheses than the orthodox theory, it can accommodate itself all the easier to a wider field of varying conditions.

As I said above, I don’t know if Keen has already done something different with his model. But so far as I can see he might well be better off with the above framework as it is extremely flexible both in the phenomena that it can explain and in the manner in which it can be used to simulate economic dynamics.

I think that Keen will appreciate this to an even greater extent if he considers the expanded version of this framework which includes a novel insight that I think has very important implications for Minskyian analysis. This insight I call the ‘paradox of speculative profits’ and I think goes a long way to explaining why financial fragility can become so acute while investors remain entirely oblivious. (Hint: this problem is built into the structure of asset markets just as the ‘paradox of thrift’ is built into the macroeconomy — i.e. it is structural). I will leave that insight, however, to emerge into the light of day when the paper finally appears in full. If Steve would like a look at the paper prior to its coming out I would be more than happy to send it to him.

Addendum: I would imagine that some commenters are going to jump on me here. “But Phil,” they will say, “a good part of this blog is taken up with your critiques of abstract modelling, why here are you seeming to promote it?” To this I would give two responses that are inherently linked.

First of all, I do not disagree with modelling per se. Rather I disagree with applied modelling as is done when models are applied to data for forecasting and so forth using Bayesian or objectivist probability methods (i.e. using econometric techniques). I have no problem with using modelling as a didactic tool provided that students are made to understand clearly that modelling and actual applied economic thinking are as different from one other as are dressing up and playing fireman as a five year old and actually working as a fireman.

Secondly, and tied to this, I think that the most valuable aspects of models are their components. If you dig into almost any worthwhile macroeconomic model, for example, you will find a Keynesian multiplier equation. I propose that this component is actually more important than the model itself in that it teaches something absolutely concrete about a very real economic relationship whereas the model likely drifts off into abstractions. What gives models their value is that they impart knowledge of these components to those who study them. In this regard I fully agree with Keynes who wrote in Chapter 21 of the General Theory,

The object of our analysis is, not to provide a machine, or method of blind manipulation, which will furnish an infallible answer, but to provide ourselves with an organised and orderly method of thinking out particular problems.

It is thus, to my mind, far more important to get the various components of a model right than it is to construct a model. But if others prefer to construct models I see it as my job merely to try to give them the correct components. Recall that it was the microchip that gave such awesome power to the modern day computer and not vice versa.

About pilkingtonphil

Philip Pilkington is a macroeconomist and investment professional. Writing about all things macro and investment. Views my own.You can follow him on Twitter at @philippilk.
This entry was posted in Economic Theory. Bookmark the permalink.

11 Responses to Steve Keen’s AS-AD Curves and a Suggestion For a New Stock-Flow Equilibrium Approach

  1. Ramanan says:

    “but we cannot avoid the simple fact that the AS-AD model is not compatible with Post-Keynesian endogenous money theory.”

    Ridiculous non-sense.

    The error of orthodox economics is not AD/AS itself but things such as how AD and AS are brought into equivalence. Neoclassical economists use “price clearing” as a mechanism and Post-Keynesians use quantity adjustments. In fact they aren’t equal in products markets and differ by “change in inventories”.

    “I would also say that the AS-AD framework does not properly incorporate expectations.”

    Translation: I am too lazy to look into economic modeling literature.

    I am not sure what your new “price theory” is but it seems to mix concepts of price determination in financial markets and in goods and services markets. Not that these things are completely independent but the equations you write are completely meaningless.

    And about your equation: price elasticity of demand measures the response of demand for something because of the change in price of it. But you have it completely opposite – you have something which affects prices because of a change in demand for it – till now but seems to have changed in your latest post. Not sure why delta is price elasticity of demand.

    • Very confused post, Ramanan. Poor show.

      The AD-curve not based on the LM-curve? Take a class in macro…

      I’m not responding to the rest because I think you’re talking rubbish…

      Email me if you want to discuss my theory or deal with it here seriously.

      • Ramanan says:

        The fact that the man in the short video uses IS/LM *does not mean* each and everything he says is wrong.

        The SFC models are just that AD/AS models in a sequential way. The right reply to the person you are addressing this on Lord Keynes’ blog is that the neoclassical price clearing mechanism in goods and services markets is wrong not that AD/AS itself is wrong.

        That still leaves the question of how prices are determined so one needs to add price theory to the AD/AS framework which SFC people do. There are two types of prices at least – one in the goods/services markets and the other in financial markets. In financial markets price clearing is not wrong but one needs more things to describe it as well.

        But the underlying message of your post that AD/AS is quite wrong is totally inaccurate.

      • How do you derive your AD curves Ramanan?

      • Ramanan says:

        I dislike all sorts of diagrams (as opposed to charts) people draw in economics and the various curves. It holds other things constant when some curve is being moved but that doesn’t happen in real life. So I find them misleading.

        The AD/AS framework is naturally suited for description of the sequence of income/expenditure/output rather than being used as some sort of curve which determines prices. Price theory is somewhat tangential to the sequence – although important.

        While most neoclassical texts use it to determine prices, some do use it for not doing it but then leave it at that. (Can’t find which one but one of the popular ones – not Mankiw/Krugman/Samuelson)

      • Price is somewhat tangential to the AS-AD? Really? Okay, Ramanan… I guess its only one axis after all…

  2. jown says:

    Very excited for when your ‘paper’ sees the light of day …

    Looking forward to the ‘insight’ … thank you

  3. Steve Keen says:

    Dear Phil,

    There’s a segment in the Hitchhikers Guide to the Galaxy where Arthur Dent orders a tea from a hyper-intelligent dispensing machine, and gets given something that is almost, but not quite, completely unlike tea.

    This post almost, but not quite, completely misrepresents my approach to economics. No offense taken, but if you’re going to write a blog post with my name in the title, at least read the relevant paper next time.

    I think That particular line comes from a lead up to my Economics E-journal paper (aspects of which, especially not using double entry bookkeeping, I am now critical of) in which I was trying to explain how monetary profits were possible when much of the Circuit literature asserted they were not–in a stock-flow confusion. I needed a pricing equation for that and its was derived from that statement.

    It then turned out to be identical to Kalecki’s pricing equation. And if you did know my work then you’d realize that I immediately put it into a differential equation as a dynamic first order time lag condition, and that while this converged to equilibrium in the simple linear model in “solving the paradox of monetary profits”, in a fuller dynamic model convergence to equilibrium does not happen.

    I don’t often find myself agreeing with Ramanan, but here I do: it’s the dynamic statement that matters, and curves don’t cut it. You won’t find a single static diagram in any of my models, let alone an AS-AD diagram.

    If you do want to check my arguments at all, this paper is a better starting point–though thanks to Matheus Grasselli, I’m now aware of errors in the final model there that I am now working on:

    I’ll finish on my opening point. One of the reasons economics has become the dogs breakfast that it is today is because of poor scholarship.

    Hicks’s original IS-LM is a good instance; though Krugman is correct that passages in Keynes can be interpreted in this way, PKs should know that IS-LM was actually Hicks’s “bread economy” model that pre-dated his exposure to Keynes.

    Let’s not repeat those errors of poor scholarship ourselves. We’ll never “fix the economists” if we do.

    Cheers, Steve Keen

    • Steve,

      I’m sorry if you feel like I misrepresented your argument. I said multiple times in the piece that I did not know if you had since gone in a different direction which apparently you have.

      Let’s not have this turn into a “he said, she said” internet argument. I’m actually interested in what you’re doing and hope that any criticisms I make are constructive. So, tell me: how are you conceptualising prices at a macro-level? Are you conceiving them as fixed prices? Are they demand-driven?

      I can’t access the link that you sent me.


  4. Steve Keen says:

    Looks like a previous attempt at a comment went awry. I agree that it’s silly to do the “he said/she said” thing, so I’ll just note that I last used AS-AD diagrams in a student essay in 1972 and get on with it.

    The point I didn’t make clear enough in the first comment is that when I derived an equilibrium price condition from equating the flow of monetary demand to the flow of physical supply in a genuinely dynamic model was that the equation that popped out was IDENTICAL, formally, with Kalecki’s markup equation:

    Pe = 1/(1-s) * W/a

    Where s is the capitalist share of output, W the money wage and a is labor productivity.

    So for this reason I have become less fussed about the cost-plus vs demand&supply driven approaches to prices. Once one drops the nonsense of marginal cost pricing, the two are identical In equilibrium.

    I then embed this in a first order time lag for a dynamic pricing equation in a system of coupled ordinary differential equations:

    dP/dt = -1/tauP * (P – 1/(1-s) * W/a)

    The fullest system there is a 14 equation monetary model of Minsky, but as Matheus Grasselli pointed out to me, it’s internally inconsistent without a variable capacity utilization function driven by the change in stocks. Adding that is the research objective for the first half of next year.

  5. Steve Keen says:

    PS maybe this link will work for that paper: it is free access (I forked out $3000 to ensure that):


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s