We should not teach the Keynesian Cross in Intro Macro

I have taught Intro Economics maybe 30 times in total. (I started very young).

Forget about the A students, because there aren't many of them. (We don't all have massive grade inflation.) Think about the C students. And forget about the Economics majors, because other people will teach them after you. Think about the students who will only take this one course in economics. And think about what they will remember, not for the final exam, but 10 or 20 years after you have taught them. What will they remember from intro macro, if they remember anything at all? Not much.The best you can hope for is a crude and fuzzy caricature.

Imagine asking such an ex-student a simple question: "what should the government do, macro-wise, to make people better off?"

The student who has been taught AS-AD will probably reply: "Increase supply? Increase demand? I'm not sure which and I'm not sure how. And should we think about inflation too? Dunno."

The student who has been taught the Keynesian Cross will probably reply: "Increase government spending!"

The first answer is nearly always a much better answer.

The Keynesian Cross should be put on The Index. It should only be taught to PhD students. But it should be compulsory for PhD students. It contains a very important insight about deviation-amplifying positive feedback mechanisms when people are quantity-constrained and use adaptive learning. And that insight is missing from New Keynesian models.

One of the reasons I really like Greg Mankiw's intro text (I was a Canadian co-author on the first 3 editions) is that it leaves out the Keynesian Cross (though it does mention the multiplier process as an aside).

My little contribution to the recent Keynesian Cross Kerfuffle.

Comments

[Livio Di Matteo]

Nick:
Have not taught Intro in a couple of years but last time did use the K-C. However, it was part of the build-up into constructing the AD-AS framework once prices varied. In other words, the students got both with the point made that the AD-AS model was a better framework in that it had both the aggregate demand and aggregate supply sides. In retrospect, I wonder if it was simply too much material thrown at them. Will re-examine the next time I teach it. Thanks for the post.

[Nick Rowe]

Livio: Yep. Nowadays, almost anyone who teaches the KC will also, I think, teach AS-AD as well, and explain how they fit together. But will the C students get the idea that the KC should only be understood as part of the background to the AD curve in AS-AD, and will they remember that subtle point 10 or 20 years later? Especially if the have the KC drilled into them via arithmetic and graphical exercises. Or will they just remember: "G up causes Y up"?

[Alex1]

What always confuses me about the Keynesian Cross is that, in it simplest version, it seems to require neither savings nor money and so it seems to suggest that you can have a demand constrained barter economy. But hidden in its dynamics are both the existence of forced savings (because if I remember correctly the equibrilating mechanism is an inventory cycle) and by extension the existence of money because I dont think its possible to accumulate inventories in aggregate unless there is excess demand for money (?). So, as far as I'm concerned the model is super misleading.
But I still think understanding the dynamic version of it is useful to deciphering the difference between the various more Keynesian ad more "classical" formulations of IS-LM.

[Luis Enrique]

"It contains a very important insight about deviation-amplifying positive feedback mechanisms when people are quantity-constrained and use adaptive learning"

update! and that insight is ....

[Nick Rowe]

Alex1: Yep. It implicitly assumes monetary exchange (it is nonsensical in a barter economy) but money is not there in the model.

I hate how it is always taught with involuntary inventory accumulation/decumulation as the equilibrating mechanism. This totally obfuscates: the fact that it works just as well in an economy where we use money to buy haircuts/manicures/massages which can't be stored in inventory; the fact that the equilibrating mechanism is adaptive learning of expected income (if instead we just jumped to the Rational Expectations/Nash Equilibrium we wouldn't see this process in action at all).

Again, these are really subtle points that are best taught to PhD students.

[Nick Rowe]

Luis Enrique: quick answer (must do admin work):

1. Quantities demanded and supplied depend not just on preferences, technology, prices, and endowments, but also on expectations of realised sales and purchases, if markets don't clear continuously.

2. My quantities demanded and supplied may be a positive function of my expected realised sales and purchases.

3. My realised sales and purchases will depend on your demands and supplies. (So we get a positive feedback loop).

4. Equilibrium doesn't just solve itself, but people must hunt for that equilibrium by updating their expectations of realised sales and purchases in the light of their actual realised sales and purchases.

That's PhD level stuff, not first year.

[Alex1]

"the fact that it works just as well in an economy where we use money to buy haircuts/manicures/massages which can't be stored in inventory"

That never occured to me before. But thinking about it, teaching inventories does seem really anachronistic in the Googlenomics/Just in Time economy. But maybe the rationale is that inventories give something quantifiable which one can easily throw into national accounts (Y=C+Inv), whereas for the haircut economy it to me seems much harder (Y=C + "forced idleness" (?)).
But I agree that it obscures (it certainly confused me) and it doesnt help that most textbooks I read in class dont even seem to try to explain the workings behind the Cross beyond "take the derivative of the formula"

[Robert Hurley]

Am I missing something here? Shouldn't the answer to the question depend of the state of the economy?

[Nick Rowe]

Alex1: haircuts force you to think about what happens when realised sales are greater or less than expected sales, and how people revise their expectations ("was this temporary or permanent?") and how they revise their planned purchases in response to revised expectations.

[Nick Rowe]

Robert: Yes, it depends on the state of the economy. But will the C students get that subtle point, especially 10 years later. Because it's not there in the model.

[Frank Restly]

Nick,

"Imagine asking such an ex-student a simple question: what should the government do, macro-wise, to make people better off?"

I would answer, regulate the after tax time cost of it's money to promote real economic growth.

[Jon]

Nick, interesting PoV. The kc is an introductory tool itself though. Intro classes need to eschew history and always try to teach the best (simplified), contemporary view. So yes, kc belongs in a history of economic thought class. Although you seem to making a broader point that it should be taught in the context of creating simplifying models as intuitive and pedological tools. Now that sure is something future professors ought to be taught!

[Linda]

Nick: your first full paragraph should be enshrined by every undergrad curriculum committee, for intro micro as well as macro!

[Paul Friesen]

Perhaps I have a somewhat unique perspective on this, because I was, at one time, one of those students you talk about who is never going to study economics after an intro course. I was a science guy. Economics was a boring, unnecessary, required course. I guess I must have passed, but can remember little.

Later in life, I did become interested. And one thing I realized I didn't understand was how the economy manages to keep most people employed. To use terms I learned later, why does aggregate supply approximately equal aggregate demand? What is the control mechanism? I must not have received an answer to this all those years ago, or at least I did not receive one that made sense to me and stayed with me.

After thinking about that quite a bit, I came up with my own answer. It's nothing new. It amounts to Keynesianism. But it's still the way of thinking about this that seems clearest and simplest to me. If a first-year student asked me for an explanation, this is what I would tell them.

Think about the money supply. Roughly speaking, that's just the sum of all the money that all the people and all the businesses have in their bank accounts and pockets. Suppose that the economy is not creating as many goods as people want to buy. What will happen?

People will be prepared to offer more money for goods. Prices will rise. If the money supply is held constant, the purchasing power of that money supply will go down, so all the people and businesses in the country will feel a bit poorer, on average, and be less inclined to buy. So prices will just rise until the amount of goods demanded equals the supply.

This can also work the other way. If demand is too low, prices will tend to fall, meaning that the purchasing power of the money supply rises. This will, again, bring the economy back to the point where supply equals demand. But there are problems with adjustment in this direction which mean that it can take a very long time. We will talk about those problems in the next lecture.

I think that's an answer I would have found convincing and reasonable all those years ago.

[Joseph]

I was a recent economics/physics undergraduate at a school with a lot of students who took economics as a second major because it either looked good or had a passing interest in it (It was almost trivially easy to add economics as a second major). In the upper level classes, even the economics majors completely misunderstood the subtleties of the Keynesian model. In a public finance course a student suggested that social security payments in the U.S. should be increased because that would lead a to multiplier effect through inceased 'G' leading to a permanent increase in 'Y'. This was after the student spending about a month on the Keynesian model in intermediate macro using Makiw's book. The sheer misunderstanding of the theory ranging from short vs. long term and what 'G' is in the model kinda floored me and I could tell that other students agreed with his point. The professor just kind of brushed it off because I don't think he fully understood how deeply the student misunderstood what he was saying and the error was in content from a different class.

I'm an 'A' student going on to an economics graduate program and I remember reviewing the Keynesian cross model section at least 10 times in Makiw's book trying to really understand what was going on beyond just a simple mechanical way. In general I find that I can follow NK economics papers quite easy compared to learning about older Keynesian models. The math is harder but conceptually a lot of the mechanisms are simpler.

[PeterN]

"That's PhD level stuff, not first year."

In detail, perhaps, but it's still necessary. Imagine a first year physics course restricted to classical mechanics, thermodynamics and classical electromagnetism. It would certainly be easier to teach and understand, but you won't find many physicists who would believe it was good idea.

Say: "Economics is filled with counter-factual simplifying assumptions. There is some disagreement on the limits that these impose on the range of validity of the models, and this is a subject of active research.

These assumptions are necessary to make the models tractable, and in more advanced economics they are often qualified in various ways, so you can't naively apply first year economics to every real economic situation." Give a few examples and leave it at that.

The K-C by itself is useless. Better Keynes would be the difference between risk and uncertainty and how expectations and experience can both influence economic behavior.

Extra points for you if you can include tail-risk.

[Frank Restly]

Robert,

"Am I missing something here? Shouldn't the answer to the question depend of the state of the economy?"

No it should depend on the desired state of the economy.

[Edward Lambert]

I incorporate the view of social capital to understand the KC.
Here is a link to a post by Michael Pettis with a graph of optimal investment against social capital in China...
http://www.mpettis.com/2013/06/10/how-much-investment-is-optimal/

The idea is that investment has an optimal level in relation to social capital. That if you tried to increase investment beyond its optimal level, social capital would not absorb it in the long-term. The consumption portion of the aggregate demand curve in the KC would be sticky in response to investment. Markets won't clear well over time. In effect the KC will not show that there is a bubble. Eventually the market will bring values of investment back in line with what social capital is truly able to absorb. The KC does not reveal what level of investment is optimal for society.

[Nick Rowe]

Paul: " And one thing I realized I didn't understand was how the economy manages to keep most people employed. To use terms I learned later, why does aggregate supply approximately equal aggregate demand?"

That's beautiful! Because that's exactly what I wondered, after I had learned the KC in high school. "OK, this theory says that, unless governments are actively controlling the economy, unemployment could be anywhere between 0% and 100%. So how come it seems to spend most of it's time closer to 5% than 50%????" And for anyone who is an Old or New Keynesian, and who knows even the tiniest bit of economic history, that ought to be a very big puzzle! my old post.

Joseph: I agree. I remember thinking about the Keynesian Cross model time and time again, trying to figure it out. It wasn't till Peter Howitt's advanced macro at the PhD level that I sort of got it straight. And I still re-think it. Very few people understand it. I'm arrogant enough to think that I understand it a helluva lot better than most economists, who often don't get it at all, but also really concerned I'm missing something.

[rsj]

The student who says "Increase government spending" will be right. What is the aversion to teaching students how the economy really works, rather than misleading them with a bunch of incorrect equilibrium models?

[Joseph]

rsj: In the short run the effect is ambiguous depending on how the spending is funded and the central bank reaction function but in the long run there is no effect, increased spending will crowd out private spending in general unless the spending is on some kind of capital good. The government spending the student was referencing was also transfer payments. In the Keynesian model, 'G' is usually defined as spending on actual goods and services and specifically excludes transfer payments.

[Joseph]

rsj: In the AS/AD, the Long Run AS curve is vertical. Government spending can't change the long run equilibrium of GDP without acting on the supply side somehow.

[primedprimate]

On a pedagogical rather than a macroeconomics note, who do you all (Nick, Frances, and the other bloggers/commenters) target while designing syllabi, lectures, tests, etc. - the A students, the B students, or the C students?

[rsj]

Joseph,

I understand the model, but the model is wrong.

Less forcefully, think of every model as being useful in some circumstances. Then the set of circumstances in which any kind of GEQ model is useful is of measure zero, whereas the set of circumstances in which the keynesian cross is useful is of positive measure.

I am not alone in this belief. Many have pointed out that DSGE models are no better than spreadsheet style accounting models for predicting inflation or growth (e.g. see Noah Smith)

And there are others:

http://rjwaldmann.blogspot.com/2012_04_01_archive.html

"But it still seems to me that you are thinking of a model in which real interest rates have an economically important effect on consumption.  I think there is essential no evidence of any such effect (last I heard GMM estimates of the aggregate intertemporal elasticity of substitution were negative.  In any case, empirical estimates of it are tiny."

http://rjwaldmann.blogspot.com/2010/05/mark-thoma-regrets-commenting-on-blog.html

"Speaking for myself only, I don't care what modern macro models say. I don't think that New Keynesian models add anything much of value to the Keynesian cross."

That's why we should just teach the Keynesian cross until DSGE models become better than spreadsheet models. If this occurs, we can teach DSGE models.

[Kevin Donoghue]

If you can't stand the Keynesian Cross, teach some variant of the world's smallest macromodel:

http://web.mit.edu/krugman/www/MINIMAC.html

[rsj]

Can we embed?

[Gene Callahan]

"I hate how it is always taught with involuntary inventory accumulation/decumulation as the equilibrating mechanism."

I don't know Nick: I found this easy to convey, especially with the recent downturn: houses and condos sit on the market. And it is easier for Macro I students to picture this than mismatched income expectations!

[Gene Callahan]

" Intro classes need to eschew history and always try to teach the best (simplified), contemporary view."

I do not agree. The best way to learn a subject is its history.

[Frances Woolley]

Primedprimate: "who do you all (Nick, Frances, and the other bloggers/commenters) target while designing syllabi, lectures, tests, etc. - the A students, the B students, or the C students?"

A good test should have questions for everyone - that is, questions that only A students will get, questions that both A and B students will get, questions that A,B and C students will get. Otherwise it's hard to separate out students. Lectures are harder to get right, because there's the risk of boring the students who pick things up quickly, or losing the students to pick things up more slowly. I try to teach to the middle and say to other people "come see me in office hours or after class," but am not sure if I always succeed. Powerpoint can be useful for quickly flashing up something that retain people's interest.

[Declan]

"What will they remember from intro macro, if they remember anything at all? Not much.The best you can hope for is a crude and fuzzy caricature."

My strongest recollection was the prof informing us that child labour was a good thing.

At any rate, on the topic at hand, I agree with rsj, as I tend to do when the topic isn't cyclists.

[W. Peden]

Declan,

You apparentely got a broad intro macro course.

[Jacques René Giguère]

My students will not be ( bar a couple) professionnal economists. They will be in business or at the Chamber of Commerce. They need to see that "necessary austerity" and "we can't spend our way to prosperity" is hogwash. ANd the KC can help me on that.
Immunologists can discuss the exquisite details of better antibiotics, Meanwhile people must know to wash their hands.

[pe]

no 45-degree. Agreed. And no AD/AS. this is even worse.

Instead teach Colander (adds a co-ordination parameter to agg p.f.) Maybe some John Bryant QJE 1983 (stag hunt) and Clower/Howitt 2000. Talk about the coordination problem and how it goes awry because of uncertainty about future. Coordination is not a free lunch.

[Mike Smitka]

I was a bit surprised that no one questions the time wasted in class trying to explain the 45º curve and how / why it works. Much simpler to either to (i) do a couple arithmetic iterations and as an add-on for better students (ii) do the algebra [which opens the way to doing a version with taxes and imports].

Now I am at an elite liberal arts college where students have all entered with good math test scores and virtually all have had calculus. I basically have no "C" students, other than those who choose not to study. [BTW that wasn't true of my institution 20 years ago.] But for most calculus was as a senior in high school, and by winter of freshman year (and certainly by winter of sophomore year), with no use of math in the interim, even this taxes their abilities. Oh, and intro economics is the largest course on campus, over 2/3rds of all students take intro micro. Given current graduation requirements (fewer social science credits required) fewer take macro. We don't offer a one-term version.

Back to pedagogy: by avoiding class time explaining the "cross" I have freed time to emphasize that the multiplier framework is a pure demand-side approach that even then is simplified. However, it provides a starting point for adding realism – various components of "it depends". Oh, and I don't use "G" as the core of the story, I will often use "I" [Canadians could use "X" with an indifferent Uncle Sam as the impulse...]. Now I avoid calling this a "Keynesian" model – I call it a pure demand-side model. I pair it with a graphical version of the Solow growth models for a pure supply-side model. I do a bit with a simple graphical life-cycle model to emphasize the (modest) short-run shocks can be muted (ditto when we discuss monetary policy). Some of this is potentially challenging material, but in the end I try to convey a message that there are very simple models, and then there are simple models, and that even the simple models get complicated very fast. At the end (along with comparative advantage, and a bit of sophistication in looking at data for (un)employment, inflation and GDP) I hope that they have a sense of the simplistic supply and demand side models, that they are simplistic models, and that they should suspect anything that sounds ... simple.

I have never found a "Principles" book I like, at least for the macro section. Most still follow Samuelson in structure – things international tacked on as an afterthought, implicitly empiricially irrelevant [which was true for the US when PS penned his first edition].

[Nick Rowe]

Back from canoeing in La Verendrye. 4 days, 3 nights, 87 kms, including 5.5 kms portages. Bugs a problem, but having a great view of the Milky Way more than compensated. I wonder how many people nowadays have actually seen the Milky Way, given the amount of light pollution?

rsj: "The student who says "Increase government spending" will be right."

Hmmmmmmm. Does that mean the optimal amount of government spending is infinite?

primed: "On a pedagogical rather than a macroeconomics note, who do you all (Nick, Frances, and the other bloggers/commenters) target while designing syllabi, lectures, tests, etc. - the A students, the B students, or the C students?"

Good question. Regardless of what we say we try to do, and what we should do, I think most of us end up teaching towards the A++ students. We are biased that way. In Intro Economics, it's especially difficult, because there is such a wide range of student abilities in understanding economics. I agree with Frances' answer (except I don't like PowerPoint, which I claim is for pedagogical reasons, but might just be because I'm old-fashioned and can't figure it out).

Kevin: "If you can't stand the Keynesian Cross, teach some variant of the world's smallest macromodel"

I find that idea very attractive. Hmm. I wonder.

Gene: If it's easy to teach the inventory story, that's because it ducks the point. If people had perfect foresight, the model would jump instantly to the new rational expectations equilibrium, where AE=Y, regardless of whether or not all goods were services.

[Nick Rowe]

Jacques Rene: OK, there is something to be said for teaching students whatever is the exact opposite of the views they come in with. To be provocative, and to open their minds to another way of viewing the world. But if your students were all Keynesians when they started your class, would you deliberately teach them classical models? Perhaps you would. There is a danger though. If you teach them an extreme model, (in either direction) they might just say "Oh well, this proves that all macro is bunk, because this model says: 'we should always increase G/recessions never happen'".

pe: interesting ideas. One thing to bear in mind though: Intro economics course content is a very (small c) conservative thing. Mankiw was a very big change, partly because it left out the KC. As Mike Smitka says: "Most still follow Samuelson in structure". It is very hard for one individual instructor or university to change from what everyone else is doing. Because it is the prerequisite for everything else in economics, and students move from one university/college to another. The government always wants us to make courses "transferable", but that stifles innovation. Plus, Intro does tend to get taught by a lot of old guys and junior instructors who don't really have the authority to try something totally different. That whole question deserves a whole post.

Mike Smitka: "I was a bit surprised that no one questions the time wasted in class trying to explain the 45º curve and how / why it works. Much simpler to either to (i) do a couple arithmetic iterations and as an add-on for better students (ii) do the algebra [which opens the way to doing a version with taxes and imports]."

I disagree. Any mathematically competent student can solve the model, but do they have a clue about understanding it?

[Nathanael]

Ugh.

I was one of those students who took 3 econ courses in college. Actually a math major, but I was pursuing a liberal arts curriculum. I was an A+ student.

I didn't think any of the courses were very good, but Intro Macro and Intro Micro were particularly bad. The Keynesian Cross is the *least* of it. The focus was just waaaaaay off, and it's way off in almost all economics course. There was no proper explanation of deflation. Distributional issues were touched on but not explained properly -- I had to look elsewhere to find the declining marginal utility of money, which is critically important to understanding the effect of distribution on the macroeconomy.

Worst of all, all economics professors draw half their graphs backwards, with the independent variable on the wrong axis. Supply and demand graphs are backwards in every single book. If you want to teach anything, you need to cut it out and put price on the x-axis.

This one thing causes over half the students to not understand what's going on. Most of them pretended to understand what was going on -- but they didn't. I kept asking more and more persistent questions for *weeks* until I finally had the "aha" moment when I said "Wait -- your graphs are all BACKWARDS? You mean price is the independent variable? Well, that makes this stuff really simple. By the way, what the hell is wrong with you people? Every other subject at this entire university puts the independent variable on the X axis, and so do you guys when you do time-series."

Intro to Economic Game Theory was better, getting into some of the basics of culturally defined fairness standards and the dominance of either Tit for Tat or Massive Retaliation strategies in repeated games. Although I was designing new experiments before the end of the course...