Come on Banks, take notice of the AS-AD-ED Model
From feedback, I need to write a better post about the AS-ED model. Thanks for the feedback from Coberly, Arne and David at CEPR. (link to previous post)
This post presents what could be a huge breakthrough in understanding the business cycle.
Banks and central banks should take notice here.
Model of Effective Demand
Figure 1. This is a model for Aggregate supply, aggregate demand and effective demand.
People are familiar with aggregate supply and aggregate demand. In the AS-AD model, AS and AD always cross at the current real GDP and current core CPI, as shown at the red dot in the model. That red dot crossing point will move horizontally to the right as real GDP grows at a stable core inflation rate. Therefore, the AS curve in the model is horizontal to represent growing real GDP at a stable inflation.
What is the Effective demand limit curve doing in the model?
Keynes described effective demand as the crossing point of aggregate demand and aggregate supply where aggregate profits are maximized. So as aggregate supply (real GDP) grows to the right, there comes a point where aggregate demand equals effective demand. In figure 1, that point is modeled around $16,9 trillion. There profits would peak and the business cycle would begin a phase of deterioration into an economic contraction, unless other dynamics counteract the effects.
What is the Basic Model for Effective Demand?
Figure 2. Basic Model of Effective Demand upon Production.
In figure 2, the upsloping straight gray line is the AS curve from figure 1 related to utilization of labor and capital. As labor and capital get more utilized, real GDP production increases. In figure 3, actual data shows that real GDP does move along this straight line from the origin of x and y axes. (2ndQ 2010 to present, quarterly) This pattern is seen in other business cycles too.
Figure 3. Real GDP grows in line with (capacity utilization * (1 – unemployment rate))
In figure 2, the curving upsloping line is the effective demand limit curve. I formulated its equation from predator-prey dynamics in Population Ecology (link) and the work of Samuel Bowles on Lenders and Borrowers under Inequality. (Bowles, Samuel. The new economics of inequality and redistribution. Cambridge University Press, 2012. pp. 42-50)
In figure 2, where the effective demand curve crosses the production curve at the stable equilibrium is the effective demand limit.
The equation for the Effective Demand curve in figure 2 is…
Effective demand limit upon real GDP = rGDP*e*T/L* (1 – (1 – 1/e)*T/L)
rGDP = real GDP
T = capacity utilization * (1 – unemployment rate)
L = effective demand limit function (labor share index * 0.76)
e = 3
The peak of the profit cycle of production can be forecasted with this model since real GDP moves in a linear path and the effective demand limit is projected onto that path. Real GDP moves toward the projected effective demand limit.
How did the model do during this business cycle?
Figure 4. 6 years of Effective Demand limit curves show great consistency. (25 quarters!) (Refer back to figure 1 for the ED curve in the AS-AD-ED model.
In figure 4, ALL, no cherry-picking, all the effective demand limit curves for 6 years crossed the AS curve in a tight zone. (see oval in figure 4) As real GDP started moving to the right on the AS curve from $14.5 trillion at the end of 2008, the effective demand curves were waiting around $16.0 trillion. As a confirmation, when real GDP hit about $16.1 trillion, the effects of the effective demand limit appeared. The ED crossing points along the horizontal AS curve have a standard deviation of $130 billion, Not too bad!
The model was a complete success in retrospect.
The equation for the Effective Demand limit curves in figure 4 is the same equation given for the ED curve in figure 2 but with L (effective demand limit function) replaced with (unit labor costs/(1+core CPI))… Now the equation can be plotted in AS-AD-ED space with core CPI on the y-axis.
- L = U/(1+C)
- U = L*((1+C)… This is how U is calculated for the equation below, since L and C are given in the model.
Effective demand limit upon real GDP = rGDP*e*T/(U/(1+C))* (1 – (1 – 1/e)*T/(U/(1+C)))
U = unit labor costs
C = core CPI %
The tightness of the zone where the ED curves crosses the AS curve for 6 years is significant. The tightness reflects stability of the effective demand limit.
Profits peaked when real GDP hit the zone marked with an oval in 2014 just as Keynes would have forecasted with his explanation of effective demand.
The peak of the profit cycle, which drives the economy and business cycle, began to come into view 6 years in advance!
Really Folks, let that sink in…
That is a huge breakthrough!!!
- The Effective Demand model holds up very well in reality.
- The over-riding relationship between labor share and the utilization of labor and capital is governed by profits.
- The ability to forecast the effective demand limit upon profit cycles has huge value for banks and their investment cycles.
The model above is worth lots of money to banks.
Any bank want to hire me? ¯\_(ツ)_/¯
Questions and comments below.
This seems like a great model, but for some reason I feel like I’m just taking your word for it. What makes it a great model? What are the underlying assumptions that you adopt and eschew? It lines up with the data nicely, but I have no idea why.
I would be very glad if you showed us more explicit work in a third post. Of course, explicit does not equate to mathy. More math could be good or bad, but math isn’t what I want to see here, I’d prefer more clarity. Indeed, one problem I find is how you’re presenting the empirics .
I applaud your effort and hope to see more.
I deleted a sentence before I posted… “The one question others always has is how labor share determines effective demand.”
That is your question. What are the dynamics behind how labor share controls utilization of labor and capital.
The first dynamic is that a lower labor share leads to a lower optimal level of capital utilization. Is that not obvious?
The second dynamic is that firms in the aggregate optimize the use of capital in every business cycle. But they do not optimize the use of labor. When capital is optimized, the effective demand limit has been reached.
But there is more…
To reiterate my confusion from the prior post, but this time referencing the AS-AD curves:
During a business cycle population is growing and people are inventing new ways of doing things, so AD increases independent of capital optimization. Each year, the AD curve should shift to the right; the AS also shifts right as firms bring on line new facilities, but you cannot see it because (in a stable inflation regime) it shifts on top of itself. These things will happen even though the things which cause business cycles will have larger magnitude.
You are telling me that the factors that leave me confused are not needed in your model. At least they have not shown up in 25 quarters of real data. Is that because they are already included somehow that I don’t see? Or because they really are negligible?
yes, population grows. Capital accumulation grows. Real GDP keeps growing. That is the nature of capitalism.
Every time real GDP shifts right, AD shifts right with it in the same spot. It is kind of a boring relationship.
Make sure that you see the difference between the AD and the ED curves. They are not the same. The AD shifts right with the AS, but the ED tends to be stable during the growth phase of a biz cycle waiting for AD and AS to catch up to it. That is how Keynes described effective demand in chapter 3 of his General Theory.
deadly words: “is that not obvious?”
Let me change your comment to Arne a little.
“The AD shifts right with the AS, but the ED tends to be stable during the growth phase of a biz cycle waiting for AD and AS to catch up to it. “
The AD shifts right with the AS, but the ED tends to be stable during the growth phase of a biz cycle. ED is the limit that the AD and AS intersection will eventually reach unless labor share is carefully maintained.
I do not think of ED as waiting, I think of it as a limit which comes into play when labor share is allowed to be reduced too much. Labor share is the income of consumers.
For 7 or 8 years my mantra has been that “Consumers can not spend what they do not have, and producers will not produce what they cannot sell.”
Obviously consumers can borrow and delay the day of reckoning but it is only delayed, and the reckoning will be more severe for it! (Charles E Persons noted that effect in November 1930.)
It seems you are basically explaining why consumer debt is at an all time high. Why rents rather than ownership is at an all time high . Why wages are at an all time low and why profits and bubbles can be deceptive, misleading to the public. It is true that we are all financing (all time high) much more to buy than ever but we are no longer the #1 producer of the world nor the #1 consumer of the world. So our effective demand would naturally have to be going down as would our growth in GDP at all time low….The big 6 bank are still loaded with toxic mortgage backed securities that they do not want anybody to know or worry about… Sen.Sanders explained this in political terms , you explain it in mathematical terms but what is your margin of error?
Wiliam, The main point of the post is to show that we can forecast the end of the profit cycle years in advance. That is huge.
“Make sure that you see the difference between the AD and the ED curves. ”
I see that and I see you are onto something. Otherwise, I would not keep trying on the things that seem just not quite right.
“we can forecast the end of the profit cycle years in advance”
I am convinced you can forecast the end of the utilization optimization portion of the business cycle, but not that it is the same as the end of the business cycle. The last two recessions (clearly) ended because people suddenly realized they had continued to grow as if they had not reached the effective demand limit and were over extended. Factors beyond capital optimization are present in all of the recessions I have experienced.
We reached the limit in the mid 90s, but did not reach the end of the cycle because we had real, non-cyclical growth. I still say your model does not show that. It is either buried in the empirical numbers, or it is missing.
Non-cyclical growth leads to a vertical trajectory component in your Figure 3. If you were at the stable equilibrium of Figure 1, then it would have no horizontal component. Perhaps most of the time the cyclical behavior swamps out the non-cyclical, but since the population component is stable, it should show up if the amplitude of the cyclical is small. Currently we have very slow growth.
When you plot Figure 3 you get your line through the origin that matches part of the data. What if that is really a locus of points on a series of increasing ED lines. Empirically, it would look almost the same during the growth regime. What happens after you reach the ED limit depends on what regime you enter next. Recession leads to a rapid horizontal change. A period of stable equilibrium is a period where ED keeps increasing. We had something close to it once, so it should be included in your research.
We may be closer to non-cyclical growth now than to recession. The Fed’s actions (always suggesting increases even though not changing the rate) sacrifices labor share for damping. In a highly damped system the underlying growth won’t be swamped out by the cycles.
The figure 3 is seen more at this post. The graph there fits better with what you are saying.
The other part of this story are the attractor states that you will see in the above link. When the attractor state opens up, the plot line in figure 3 will start to go vertical like you pointed out. Then you get non-cyclical growth as you say, but it is really transitional growth from one biz cycle to another…. of course the late 90s was an exception.
Thanks for the link. I was implicitly referring to exactly that graph. I think you have a nice description with your Cobra equation of the up to the right business cycle growth and with your empirical Effective Demand of the limit of each growth cycle. I don’t think you can get a model of the leftward motion because I think different factors have dominated each of the post war depressions.
You refer to “transitional growth”, but referring to your linked chart, most cycles restart the Cobra Equation regime with almost no non-cyclical growth. The 90s and now are “exceptions”. Or they are times when different regimes dominate.
Looking at the chart, I could observe that business cycle growth dominates in the 2 to 6 year range and assert that population and productivity dominate when you get out to 20 or 30 years. I suggest that you can also see productivity dominating in the 90s and that rather than calling it an exception you can incorporate it into the model.
First attempt – just add it. That would imply that the ED limit is not fixed for the business cycle, but in the absence of rapid non-cyclical changes, it won’t change by enough to impact the cyclical growth. The data are probably too noisy to tell the difference in most cycles.
I generally distain analogies, but this is mine, so I have more tolerance. 🙂
The escalator analogy to your Effective demand research:
the motion of the escalator takes us up to a new floor as described by the Cobra equation. The new floor is at the level of ED. The escalator is flawed, because it goes past the floor and the dumps people instead of just reached the new floor. The housing boom escalator was particularly nasty because people’s insistence on continuing to ride it too high can actually have the effect of allowing them to ride it higher through the magic of speculation.
If you walk up the escalator you have not changed the locus of points which describe your path.
To torture the analogy:
1) Since the top of the escalator is not at the next floor, if you take only a single step along the way, is it possible to observe whether the floor changed its position because you took a step?
2) If you slow down the escalator, do you reduce the chance of people landing on their faces when they reach the next floor?
3) Slowing the elevator makes it easier to observe the effects of people taking steps.
4) Adding more dimensions can go from interesting to intractable in a hurry.
Effective demand is a limit like in a predator prey relationship. Let’s say that capital share actually consumes its own capital utilization by lowering labor share. By paying labor less, capital will use less of its capital.
Once capital income has optimized use of its capital resources, profits become a zero-sum game. When one firm gains profits, another somewhere will lose. Once this zero-sum game develops, aggregate profits start to decrease. That is where we are now.
Economic growth is not zero sum. Therein lies the problem I have with your model. In your analog, the prey’s range is fixed. Their food supply is limited by what can be grown in the area you are analyzing. To better represent our economy over time you would need to let the food supply increase even if the prey population is stable.