Winner of the New Statesman SPERI Prize in Political Economy 2016

Wednesday, 5 June 2013

Mystified on AS/AD

Peter Dorman asks why you don’t see AS/AD being used in the blogsphere, but it is still there at the beginning in textbooks (ht Mark Thoma). Various people have responded, and in most cases I am very puzzled. I think Paul Krugman, as ever, gets to the heart of the issue, but I still puzzled by what he says. My problem is not with any of the macroeconomics, but with what others think is easy or otherwise for students.

As Paul says, the diagram that fits more closely with how macroeconomists think has inflation and not prices on the vertical axis. The Phillips curve is one of the key relationships in Keynesian business cycle analysis. It relates the two variables that policymakers talk about. So it is just natural (and not at all the result of any fetishism) to start with a diagram relating inflation and output.

It is also very easy to give students an intuitive explanation about why there is a short run relationship between inflation and output, rather than prices and output. You just need to explain how, when output exceeds a natural rate, workers want higher real wages, but firms will not concede a lower mark-up, and may want higher profit margins. So we get a wage price spiral which in the short run gives us some particular level of inflation. The fact that the Phillips curve depends on inflation expectations, which could be endogenous in a dynamic way, is again not that difficult for students. In the long run the Phillips curve is vertical, because if output remains higher than the natural rate, expectations about inflation will rise etc etc.

The problem Paul seems to have is with the other curve, which he quite rightly explains tells us about how policy reacts. But why do we need another curve? We can think about monetary policy choosing some level for the output gap, and tell all the stories we want about the past on that basis. OK, maybe that is too much – economists are addicted to mimicking supply and demand. So if you want to derive a ‘demand curve’, just have policy choose the output gap to minimise a quadratic in excess inflation and that gap, subject to the Phillips curve. (Here is a textbook that takes this approach from the start.) This is much easier than deriving the conventional AD curve in output and prices space. [1]

Paul gives two reasons why he still likes AS/AD. First it provides a quick introduction to why demand and supply shocks are different. But can’t you do the same thing by talking about moving along the Phillips curve and shifts to the Phillips curve? Second, he says that one advantage of AS/AD is that it conveys how ultimately the economy is self-correcting through price adjustment. Now I agree it is vital to have the vertical long run Phillips curve there. But in terms of self-correction, I think the AS/AD is downright misleading. It encourages students to think that there is some kind of automatic adjustment of demand to inflation, rather than policy induced adjustment. For every good student who understands that inflation reduces demand because it leads policy makers to increase (real) interest rates, there are as many lazy students who take away from AS/AD that macro is just like the micro they have already learnt, which is a fatal mistake to make.

So I cannot see that AS/AD represents something that is easier for students to understand than the Phillips curve. Instead I would call it something that it is easy for students to misunderstand. Even if they do not misunderstand, starting off with AS/AD forces them to go through an awkward transition phase to get to a place where they can understand what is currently going on. The number of times I have been asked by students ‘do I use the Phillips curve or AS/AD?’  Let us go straight to the Phillips curve, and make our and their lives easier.

[1] Nick says quite rightly that the more conventional AD curve does tell us about other monetary policy regimes besides money targeting, but of course in the long list he gives inflation targets do not appear. So I’m mystified about why you start students off by telling them how things worked in the (maybe) past, or the (maybe) future, but not the present.


  1. OK, I'll admit: *sometimes* (depending on the monetary policy regime, and your theory of price stickiness) it might be more useful to put inflation rather than the price level on the vertical axis.

    Here's a fun example where putting inflation on the axis makes more sense. Or maybe a very frightening example, if you are American, or just worried about what the Fed might do next:

  2. But actually, if we are targeting 2% inflation at a 2 year horizon, and we take the current price level as predetermined, wouldn't it make perfect sense to put P on the axis, draw a horizontal AD curve at 104, (today's P=100), and an upward-sloping SRAS curve? Monetary policy has almost no influence on today's P or Pdot. It's mostly predetermined. What the BoE and BoC care about is where AS and AD will be 2 years in the future.

  3. Simon, We can combine the Phillips curve with the AS-AD model into a new model... the Aggregate supply - Effective demand model. Here is the post...

    The equation for effective demand gives the spare capacity available between real GDP and the LRAS zone. When the spare capacity is used up, the dynamics of the Phillips curve kick in.
    Through the equations for the limits of effective demand, the AS-ED model (aggregate supply - Effective demand) gives economic constraints that are missing from the AS-AD model.
    Case in point, when the crisis hit, the price level of effective demand (on the AS-ED model) shot up to almost 30%, while inflation at aggregate supply fell to 2%. There was a huge separation between the effective demand and aggregate supply curves.
    The effective demand curve behaved like a normal demand curve should. As output dropped, demand rose for that lower output. Also, for that lower inflation rate, there was more demand for output by following the inflation line out to the effective demand curve.

    What was the effective demand curve telling us that the aggregate demand curve wasn't?

    1. The effective demand curve was giving a measures of spare capacity and low input use.
    2. As output declined, interest rates declined, saving increased, leading to a decrease in the demand for liquidity. The effective demand curve was revealing unsatisfied demand, in other words... "effective demand".
    3. The effective demand curve was revealing that a price level of 30% would have been paid for such low output, since the market was not clearing appropriately with full-employment (a measure of dead-weight loss to society).
    4. The crossing point of the effective demand and the aggregate supply curves showed the equilibrium price level and output if productive capacity held constant, there were suddenly "full employment" of available labor and capital (within the constraints of effective demand) and unit labor costs rose in par with the price level.

    You can actually use the model for nominal GDP dynamics too.

    Real GDP is always on the aggregate supply curve. The full-employment constraint is on the effective demand curve. When there is excess demand (effective demand), factor inputs do not compete to cause inflation.
    The US currently shows a real GDP limit of $14.100 trillion if inflation does not change and productive capacity is increased with available labor and capital (within the limits of effective demand)... approx. $400 billion more in real GDP before spare capacity is used up and inflation pressures start to mount.

    All the above principles show that the AS-ED model is more useful than the AS-AD model for assessing economic dynamics.

    1. Here is an application of the AS-ED model (aggregate supply - Effective demand) in looking at the time period when the natural rate of unemployment increased. There was a window of time when effective demand fell rapidly. Unemployment should have fallen too, but it didn't.

  4. Simon: " there are as many lazy students who take away from AS/AD that macro is just like the micro they have already learnt, which is a fatal mistake to make."

    Yes. But they will think that macro is just like micro even if we don't show them AS-AD. The only way to deal with that problem is to face it head on. "Let's add up all the demand curves, and add up all the supply curves, and since each demand curve slopes down AD slopes down, and since each supply curve slopes up AS must slope up! Now, why is that wrong? Let me tell you why." And the number of students who think that AD slopes down because: "Duh prof! If prices go up we can't afford to buy as much stuff! So inflation makes us poorer, and also causes unemployment!"

    1. The demand curve slopes down because a higher price level implies less aggregate output demanded. What is the mechanism behind this? The strength of demand is based on labor income to buy finished goods. So let's look at labor share.

      labor share = unit labor costs/price level

      As the price level rises, the balance of labor share declines. Thus, holding all else constant, demand will decrease with a rise in the price level. It is too simple.

      The supply curve slopes up. According to my model, capacity utilization, which increases with output, is measured in relation to labor share.

      real output = potential RGDP + a * (capacity utilization - labor share)/labor share

      As price level rises, obviously holding other variables steady, labor share falls, which increases real output. Do the other variables hold steady? No... That is why the AS curve shifts right when there is spare capacity and when there is sudden jump in demand.

      But the effective demand and aggregate supply curves respond to the price level in the appropriate way.

      And Nick, my ancestry is from Quebec, Lalonde. My family goes back to one of the first 15 families in Quebec, about 500 years ago.

    2. Nick: Of course poor students will be tempted to think macro is just like micro. But we should not encourage them. If we started with the Phillips curve, they would be less inclined to think this is just a micro AS curve. If we derived our policy reaction function explicitly, and refrained from calling it a demand curve, they would be less likely to think it is just a micro demand curve.

  5. This comment has been removed by the author.

  6. Simon, I agree with nearly your entire post, but I do think there's a difference between the LRAS curve and the Phillips curve that needs to be pointed out. The LR P-curve is a locus of stationary points. It takes additional assumptions that many, including myself, would not make to invest it with equilibrium properties as well. The LRAS curve, however, *is* a potential equilibrium locus: the argument is that this is where the economy is going to go. Whether NAIRU is also an equilibrium -- whether there is an adjustment process that produces it in the long run -- is a question that goes beyond the scope of this discussion. What matters here is that proponents of AS-AD should recognize that they are putting a stronger interpretation on the LR P-curve.

    (As for AD, I prefer to keep policy outside the model on the principle that economic models should be deployed *for* policy.)

    1. Peter: The point you make contrasting the LRAS curve and the LRPC is interesting, so it sounds like a discussion worth having. On AD, I also agree, which is why I’m reluctant to automatically derive a policy reaction curve (and would certainly not call it AD). I wish I didn’t have to explain to students brought up using these conventional models that if there is a demand shock of known size and timing, policy can and should try to completely eliminate that shock so there are no consequences for prices or inflation.

  7. Peter,
    The LRAS curve is somewhat analogous to gravity and the center of the earth.
    Where does gravity pull us? To the center of the earth.
    Where is gravity the strongest? Not at the center of the earth, because we are being pulled in all directions at that point. The greatest point of gravity is actually near the surface.

    Thus, we are pulled toward the LRAS curve because there are profit incentives to employ more labor and capital. Yet, when we reach the LRAS curve, there are no more profit incentives that act as gravity pulling us to employ more labor and capital.
    The LRAS curve is not a strong equilibrium. It is a weak and vulnerable equilibrium. The equilibrium can easily be influenced by wage demands, changes in asset prices, monetary policy... because there is little counter-active force from profit momentum.

    The effective demand - aggregate supply model (AS-ED) shows where the LRAS curve is by the crossing point of effective demand and aggregate supply. The difference between the curves in terms of price level is a measure of the gravity (profit incentive) pulling the economy toward the LRAS curve. The effective demand of spare capacity goes to zero at the LRAS curve. Thus representing no profit incentive to employ more labor and capital.
    In the AS-ED model, there is a sweet spot in the measure of spare capacity between 4% and 6%, where the productive traction of profit incentive seems to be in balance with spare capacity. Productive capacity grows the best in that space. Just as gravity serves us best at the surface of the earth, some spare capacity away from the LRAS serves economic growth best.

  8. This is in as part of my service to help all we who need assistance on these topics.

    The beauty here and in other topics is that one can learn almost as much from the comments as from the article.

    Again thanks so much

  9. 1. Unlike Simon (and others), I do have a problem with the macroeconomics of the AS-AD model, and I think he (and others) should have a problem also. I am talking about the standard AS-AD model presented in textbooks – AD derived from ISLM and AS derived in a variety of ways, all of which are logically inconsistent with the ISLM theory of output.
    I posted a summary of my criticisms on Peter Dorman’s blog last week:

    2. I agree with Simon that most economists think in terms of inflation and the Phillips Curve, not the price level. And if they don’t, they should, because prices don’t fall anymore.

    3. I also agree that the AS-AD model is much more difficult to teach and understand (because of its logical inconsistencies).

    4. I also like Simon’s emphasis on the fundamental differences between the macro AS-AD model and the micro S-D model, and that the apparent similarities lead to confusion rather than understanding.

    5. When students ask: “which model should I use?”, if the assignment is to analyze the real world, the answer would have to be ISLM+PC, wouldn’t it?

    6. I can’t believe Krugman’s justification for AS-AD – that it explains the economy’s automatic adjustment mechanism. Does he really think that deflation is an unproblematic adjustment to full employment (except that it may take a few years)? From a Keynesian?

    I was interested to read that Krugman cringes every time he gets to AS-AD in revising his textbook. He has good reasons to cringe. He should just toss the model, and then he would not have to cringe and write absurdities.

    7. Simon asks, “why do we need another curve?” My answer: we don’t; and the extra curve is logically inconsistent and empirically unrealistic and confuses students. We should stop teaching the AS-AD model! I stopped teaching it 15 years ago, after reading Colander’s brilliant 1995 JEP article entitled “Stories We Tell …”.

    Fred Moseley
    Professor of Economics
    Mount Holyoke College


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time.