The Keynesian Illusion


by David K. Levine, March 2015

1. Introduction

The philosophy of Keynesianism has an intense appeal. The cure to economic ill is for the government to spend more money - and no reason to worry about how it is paid for. Not only does the Keynesian medicine cure our ills - it even tastes good going down. Austerity is a horrible mistake. The Great Recession could have been prevented if only the government spent more. It would have been infinitely worse if not for the stimulus that did take place. This is repeated by people whose opinions I respect: the jurist Richard Posner, the blogger Kevin Drum - and many others. And we all know that the underlying theory is that of John Maynard Keynes.

Let me give Richard Posner's explanation of Keynes from his New Republic article explaining how he became a Keynesian.

Income spent on consumption, in contrast to income that is saved, becomes income to the seller of the consumption good. When I buy a bottle of wine, the cost to me is income to the seller, and what he spends out of that income will be income to someone else, and so on. So the active investment that produced the income with which I bought the wine will have had a chain-reaction--what Keynes calls a 'multiplier'--effect...For Keynes, in other words, it is consumption, rather than thrift, that promotes economic growth.

From this the conclusion that in hard times the government should stimulate demand - that each dollar it spends will increase economic activity not by one dollar but by one dollar times a multiplier so that the expenditure (in effect) more than pays for itself.

There are two parts of this: the story of the multiplier and the conclusion about government spending. The former seems like common sense: we all know that if a city builds a huge sports stadium local burger flippers will benefit from increased sales - and in turn they will buy more of other things such as haircuts and tattoos thereby benefiting hairdressers and tattoo artists. That seems to support the first part of the story. Support for the second part - and the evidence that held sway over the economics profession for decades - was the Great Depression and especially the end of the Great Depression. It is difficult these days to understand just how bad the Great Depression was - how vast was the number of prosperous middle class people who were reduced to paupers barely able to feed their families. We talk as if the "Great Recession" is somehow comparable - but it does not (in the U.S.) come close. And how did the Great Depression end? With World War II - just as Keynes seemed to recommend the government engaged in massive spending financed by borrowing and printing money - and voila - the Great Depression turned into the great prosperity.

The thing is that the Keynesian prescription - spend more and don't worry about the bills coming due - sounds a bit too good to be true, almost like a perpetual motion machine.

At which point I want to refer you to a drawing by M. C. Escher. It illustrates a canal with water flowing downhill and around several corners until it reaches a waterfall which runs a waterwheel, then flows around the canal downhill back to the top.

Escher Waterfall

This is an illustration of a perpetual motion machine - which of course we know is impossible. How do we know this? If we measure the angles carefully and do the calculations we will discover - of course - that the water is flowing uphill. And we might wonder if the reason that Keynes is so popular with those that do not do math is because they cannot measure the angles carefully. We might want to subject Keynes rather vague stories to some careful measurement.

Another thing we might do with the Escher diagram is to try to build the machine - in which case we will discover it is impossible. As far as Keynes goes, we will come to that as well.

2. Keynes theory, insofar as there is one

I want to subject Keynes story to some scrutiny - to measure the angles carefully. I could do this using complicated math - and if I wanted to be at all realistic I would do just that, but it is his reasoning I am after and I can measure that with a simple example that is precise but avoids any math - it is the precision that matters.

I want to think here of a complete economy peopled by real people who produce and consume things. Let's say four of them: a phone guy who makes phones, a burger flipper, a hairdresser and a tattoo artist. Let's say that the burger flipper only wants a phone, the hairdresser only wants a burger, the tattoo artist only wants a haircut and the phone guy only wants a tattoo - around the circle in effect. We'll suppose that each can produce one phone, burger, haircut or tattoo and that each values the unit of what they want to buy more than the unit of what they can sell. That is, the hairdresser happily cuts hair if he can get a burger and so forth. What happens is clear enough: the phone guy produces a phone, trades it to the tattoo artist in exchange for a tattoo, who trades the phone to the hairdresser in exchange for a haircut, who trades it to burger flipper in exchange for a burger. All are employed, all get what they want - everyone is happy.

Now suppose that the phone guy suddenly decides he doesn't like tattoos enough to be bothered building a phone. Now the circle is broken and this is a complete catastrophe. Everyone is unemployed. Demand is insufficient. There isn't enough consumption - none at all in fact. And notice how this works: one person - the stupid phone guy who is causing the problem by not wanting to buy a tattoo - is "voluntarily unemployed" - he is lazy and doesn't want to work. The other three are "involuntarily unemployed" each one is willing to work in exchange for pay. The burger flipper would like to work making burgers if he can get a phone, the hairdresser would like cut hair if he could get a burger and the tattoo artist would like to work if he could get a haircut and yet all are unemployed.

Now to the multiplier. Suppose - instead of building a sports stadium - the government gives a phone to the phone guy. Why then he'll sell it for a tattoo (he doesn't have any use for a phone himself) the tattoo artist will use the proceeds to buy a haircut, and so around the circle. Full employment. Just put in one phone and you also get a haircut, a tattoo and a burger! That is the multiplier, and just what we see when the city government builds a sports stadium. Nothing mysterious here.

But...this what an economist would call standard competitive equilibrium theory - meaning there isn't a free lunch, and indeed we better ask - how did the government get a phone to give to the phone guy? One is reminded of the old economics joke:

A physicist, a chemist and an economist are stranded on an island, with nothing to eat. A can of soup washes ashore. The physicist says, "Let's smash the can open with a rock." The chemist says, "Let’s build a fire and heat the can first." The economist says, "Let's assume that we have a can-opener..."

Is the basis of Keynesianism that we should assume that the government has a phone to give away? Well maybe not. Maybe the government should follow Keynes advice and print some money (or bury it) and give it to the phone guy. Then the phone guy can buy a tattoo, and the tattoo guy can buy a haircut and the haircutter can buy a burger, and the burger flipper - ooops...he can't buy a phone because there are no phones. There are two possibilities. One is that the burger flipper realizes he shouldn't sell the burger because he can't buy anything he wants with the proceeds, and we are right back where we started with everyone unemployed. Or maybe he doesn't realize that and gets left holding the bag. There is a word for that kind of scheme - it is called a Ponzi scheme and sometimes they work - people do make mistakes - and sometimes they don't - and it seems like a poor excuse for economic policy that our plan is that we hope the burger flipper will be a fool and be willing to be left holding the bag.

Now there is something else we can try: we could force the phone guy to make the phone - then he might as well sell it and get the tattoo, and the world is good - for everyone except the phone guy who is forced to build a phone he'd rather not build. Now there is a sense in which that might be worth it - after all, we helped three people - the tattoo artist, the hairdresser, and the burger flipper at the expense of one - the phone guy. But then let's at least not pretend that there is a free lunch - let's be honest and say that we are screwing the phone guy to help everyone else. It doesn't seem that this is what Keynes or the Keynesians are talking about - and forcing people to work against their will probably seems pretty fantastic as an economic policy...but let's go back to the example of World War II because that is exactly what the government did - not only did they spend a lot of money that they borrowed or printed, but they also drafted soldiers into the military and forced a lot of businesses to produce and do things they would really rather not have done. And certainly while economic activity may have picked up a great deal during and after the war - it is doubtful that the draftees who died in the war benefited much from this. So the big evidence for Keynesianism fades away when we scrutinize it a bit more closely - and what did happen seems to accord pretty well with very simple and very classical economic theory.

Let's talk a bit more about how the government might get a phone - our poster child for anti-austerity at the moment is Greece. What if Germany gives a phone to Greece? That would be great - if the phone guy gets a phone, as we already know we are back to full employment - the multiplier and all that. But: what if Germany is just like Greece, except that their phone guy likes to make phones? If we take the phone away from the German phone guy then Germany collapses into unemployment - the multiplier works just as well in reverse. So it is with sports stadiums - it's great for the businesses nearby - but the money came from somewhere, and of course all the businesses that used to deal with the people that used to have that money are going to suffer. In our example it is all a wash - we can have employment in Germany or in Greece but not both - and as it is the German phone guy who is willing to build the phone it doesn't seem complicated to figure out what is going to happen.

But Keynes (and Posner) are very worried about investment and savings. So what if the reason the phone guy stopped producing phones and buying tattoos was because he wanted to spend his time creating the "next great thing" that will lead to world peace and prosperity in a few years time? We could fix the unemployment problem by forcing him to produce phones - but then of course the next great thing will never happen. There is no investment and no R&D without savings - so beware the theory the way to higher growth is through lower savings.

So there you have it: if we take Keynes theory and measure the angles carefully we discover you can't make a perpetual motion machine.

3. Can you build one?

That's all theory, and I'm sure that you are a practical person. So while unlike Keynes I can produce a precise theory that explains the facts of sports stadiums, multipliers and World War II - maybe my theory is wrong and we really can build perpetual motion machines? Well it has been tried...I quote from Richard Nixon who in response to a recession that began in 1969 said in 1971 "I am now a Keynesian in economics." As a consequence the U.S economy fell into a decade of "stagflation" with high unemployment and high inflation - something that Keynesian theory could not explain (inflation is supposed to cure the unemployment) - and something that Keynesian economists spent countless hours and professional publications trying to explain - and with so little success that besides a few die-hards the rest of the profession quite rightly abandoned the theory. We tried to build the perpetual motion machine - and it didn't work.

For those of you who - sensibly - are interested in facts - bear in mind that the main tenet of Keynesianism is that the key to growth is to avoid saving. The country that has had the most phenomenal growth - in the history of the world in fact - has been China over the last few decades. Will it surprise you to learn that the savings rate in China these decades has been nearly 50% - probably also the highest in the history of the world?

4. M. C. Escher versus J. M. Keynes

The comparison between Escher and Keynes is not especially fair to Escher who drew a simple elegant device not with the view that anybody would try to build one, but as an clever illusion. By contrast Keynes theory is neither simple nor elegant - the General Theory is a kind of Rube Goldberg device

Goldberg alarm

with many spinning wheels, tiny bridges, weights that go up and down on threads - and if carefully examined, many of the parts have no connection to any other parts. Those who have some knowledge of Keynes may object to my discussion of the multiplier above - despite the fact that it predicts all the facts about which Keynes talks - because it isn't what Keynes had in mind. Specifically, Keynes discusses the possibility that - even if the phone guy is willing to produce a phone - there might not be any trade at all because the tattoo artist is unwilling to accept a phone in payment for a tattoo for fear that the hairdresser will not accept the phone and so forth. This coordination failure equilibrium where nobody trades because nobody believes that anyone else will trade has (obviously) different policy implications than a phone guy who does not wish to produce a phone. Unfortunately it is unclear what the government can do about it: even if they give the phone guy a free phone or force him to produce a phone it doesn't seem to solve the problem.

Here is the thing: the possibility of coordination failure pretty much has nothing to do with the multiplier. It is true that if there is purely bilateral trade: the phone guy wants a tattoo, but the tattoo artist wants a phone there is unlikely to be a problem - we figure the two of them can get together and work things out. But as soon as there are more than two needed to trade the possibility of coordination failure seems a possibility. Moreover, although Keynes speaks of coordination failure, I don't believe that it is regarded as particularly "Keynesian." One the one hand it is not part of most of what are considered "Keynesian models." On the other hand it is part of virtually every modern model in which there is money - even though few would be regarded as "Keynesian." It is surely true that if money has no intrinsic value and is used for trade there is the possibility that trade will collapse because nobody expects money to have any value. But: if that is the case, the government is unlikely to solve the problem by printing additional money that nobody wants.

There is also talk in Keynes about the role of expectations in coordination failure - optimism and pessimism. For example: pessimistic people expect money to have no value, they don't trade and it is a self-fulfilling prophecy; optimistic people expect money to have value so they trade and this is also a self-fulfilling prophecy. Hence it may be that the government can fix things (if things are indeed wrong) by convincing people to be optimistic. That would be wonderful, but unfortunately the evidence suggests it doesn't work. FDR gave great speeches about "There is nothing to fear but fear itself" that greatly uplifted spirits - but didn't do anything to end the Great Depression. Or take the current crisis in Greece: with the election of the Syrzia government optimism - as measured by polls - went through the roof. If Keynes was at all correct we would expect economic activity to soar, people to get back to work, start paying their taxes, the Greek stock market to take off, and so forth. Nothing could be further from the truth: economic activity is as anemic as ever, people are as unemployed as ever, few people are paying taxes - and everyone is still trying to get their money out of the country.

Keynes General Theory is a long book - and any book with many anecdotes and ideas is bound to be right some of the time. But whatever else it is, it is neither general nor a theory - and as either a guide to economic research or practical policy is essentially useless.

5. Conclusion

The point is that Keynes is as fascinating and tempting as the Escher drawing - and makes as little practical sense. Economists have worked for decades trying to make sense of Keynes theory and use it to explain the facts about depressions, recessions, crises, unemployment and so forth. It is hardly the case that a conservative profession dismissed Keynes and refused to take him seriously, that the economics profession never gave him a fair shot. Quite the opposite: some of the most brilliant minds in the profession convinced of the absolute truth of Keynes ideas spent decades trying to make those ideas work - they and we have failed.

I do know about Keynes and about Keynesianism. I am old enough too have been taught Keynesian theory as orthodoxy both as an undergraduate and graduate student. My father - a PhD economist whose PhD advisor was the great Keynesian Nobel Prize winner James Tobin - called himself a Keynesian - almost as a statement of religion. In my youth my father and I wrote an empirical paper using a Keynesian model. I studied with Keynesians - I took classes from Bob Solow, an avowed Keynesian. I worked as a research assistant for Stan Fischer doing empirical work using a Keynesian model. One of my mentors was Axel Leijonhufvud whose great and famous work was a book called On Keynesian Economics and the Economics of Keynes. When I was a student we had to study the history of economic thought - I have read Keynes General Theory and some of Keynes lesser works - I even got a perfect score on an examination about them. I participated in (witnessed mostly) a long debate between Leijonhufvud and another great Keynesian Don Patinkin about what Keynes really meant and actually said. Knowledge of Keynesianism and Keynesian models is even deeper for the great Nobel Prize winners who pioneered modern macroeconomics - a macroeconomics with people who buy and sell things, who save and invest - Robert Lucas, Edward Prescott, and Thomas Sargent among others. They also grew up with Keynesian theory as orthodoxy - more so than I. And we rejected Keynesianism because it doesn't work not because of some aesthetic sense that the theory is insufficiently elegant.

Keynes own work consists of amusing anecdotes and misleading stories. Keynesianism as argued by people such as Paul Krugman and Brad DeLong is a theory without people either rational or irrational, a theory of graphs pulled largely out of thin air, a series of predictions that are hopelessly wrong - together with the vain hope that they can be put right if only the curves in the graphs can be twisted in the right direction. As it happens we have developed much better theories - theories that do explain many facts, theories that provide sensible policy guidance, theories that work reasonably well, theories that are not an illusion. The current versions of these theories are very unlike caricature theories of hopelessly rational people who are all identical. Current theories are not perfect - but unlike the Keynesian theory of perpetual motion machines they explain a great deal and have a great deal of truth to them. A working macroeconomist reading Krugman and DeLong feels as a doctor would if the Surgeon General got up and said that the way to cure cancer was to draw blood using leeches.

Beware of politicians saying "you never warned us" when the truth is "we ignored your warnings" - and equally beware of economists bearing empty promises of perpetual motion machines. And when it comes to government borrowing, remember that if in the long run we are all dead, hopefully our children will not be.


After I posted the essay above Steve Blough twittered to Brad DeLong about strawmen. Fortunately Brad has provided us with  a beautiful illustration that captures perfectly the problems with the type of reasoning that I was writing about - so hopefully my discussion of his rejoinder will not bring any accusations of strawmen.

I'm going to start with "Everybody else believes that the right way to model the economy is not the barter economy of DKL." My simple example is carefully constructed so that barter is impossible. If, to quote myself "It is true that if there is purely bilateral trade: the phone guy wants a tattoo" that is model of barter. But the example with the phone guy, tattoo artist, hairdresser and burger flipper is an example of an economy with commodity money: every trade is for a phone, so that the phone serves as money in this economy. So it is in a gold based economy where trades are goods against gold but some people like gold for its intrinsic value.

Brad is quite right that we do not live in a world of commodity money - although Mill and Malthus who he quotes did live in a gold based economy. And I should emphasize that the simple example I constructed was not intended as a realistic guide to an economy or to provide useful policy advice - the purpose was to examine carefully Keynesian reasoning - to, as it were, measure the angles in the Escher diagram carefully. However, there are a few more things we can do with the simple phone-tattoo-haircut-burger economy. We could, for example, have trade take place using IOUs rather than phones. So that the phone guy trades the tattoo artist an IOU for a phone in exchange for a tattoo, the tattoo artist trades the IOU for a haircut, the hairdresser trades the IOU for a burger, and then the burger flipper redeems the IOU in exchange for the phone. This is called inside money and as you can see nothing much changes - except that the role of trust is highlighted - each must trust the promise of the phone guy to give the phone in exchange for the IOU, and you can see how there can be coordination failure if trust fails - again without an obvious way in which government can restore this trust.

Now Brad has in mind not commodity money or inside money but outside money. This is trickier - which is why I avoided it in the original example. We need to start by supposing that people have already some outside money - say each has a dollar bill. More important - since there will still be four dollars floating around at the end - we need to think why they want to hold those dollar bills. I'll come back to this, but for the moment, let's just say that each person wants to hold a dollar bill at the end for some reason or other. Then the phone guy can give a dollar to the tattoo artist in exchange for a tattoo, and so the dollar bill travels around the circle until it reaches the burger flipper. At this point because the phone guy has no dollar bill and would like to have a dollar more than a phone he trades the phone to the burger flipper for the dollar bill. On the other hand, if the phone guy decides he doesn't want to make phones, giving him a second dollar doesn't seem to solve the problem - he could trade the extra dollar around the circle, but in the end he would have to exchange a phone for a dollar - which we just agreed he doesn't want to do.

Brad wants to do another conceptual experiment in my example "Suppose that the phone guy has a bunch of Mortgage-Backed Securities in her retirement portfolio, which suddenly crash in value with the financial crisis...The phone guy looks at her zero-value retirement portfolio and sets to work building a phone--but decides to use her earnings not to get a tattoo but to hold them in the form of outside money: cash." Let's ignore the bit about mortgage backed securities as there are no houses, house buyers, or brokers in the example economy. If we think these people are important, then we ought to put them in the model, explaining who they are, what they want, and whether they also might want phones and tattoos and so forth. Anyway - to try to make sense of Brad - let's stick with the simple outside money economy. The idea seems to be that the phone guy decides - because of the loss of his retirement portfolio or whatever - that he wants to have a second dollar more than a tattoo. So indeed the economy collapses. But if the government gives everyone a second dollar everything is fine - they can trade two dollars around the circle, and the phone guy will be happy because he can have a tattoo and two dollars. Think about that. How is the economy in which everyone has two dollars different than the economy in which everyone has one dollar? If this really works, can't the government solve the economic crisis just by declaring that quarters will be called dollars - in which case everyone has four times as many dollars? At this point we really need to ask who is holding money and why - and if the government gives people money, to whom do they give it? Perhaps if the government gives everyone an extra dollar - and so prices double - the phone guy will decide on account of inflation that he really wants four dollars not two for his retirement? To make sense of this we need to understand why people are holding dollars in the first place - that is why in my simple example I used commodity money: with outside money people presumably hold it not because of the beauty of the artwork but to buy things with it in the future. But to study that we need to introduce a future, and that takes use beyond what we can do in the simple Keynesian framework of a single period and beyond what can be done without math - but hardly beyond what is done in modern macroeconomic models.

So what happens if we go beyond the simple one period Keynesian framework? What if people hold outside money so that they can use it for future trades? What happens if over time people sometimes want to buy and sometimes want to sell - although they don't always know in advance which - and they hold outside money for the times in the future they will want to buy without wanting to sell? Here is a quote from the introduction to a 1992 paper that studies such a model: "a trader may occasionally find himself in a position in which he wishes to buy from another trader who wishes to sell but does not have enough money on hand to make the purchase. An expansionary [monetary] policy that gives all traders equal amounts of currency effectively redistributes wealth from the relatively rich seller to the relatively poor buyer. This can make possible socially desirable trades that would not otherwise take place." Now there are a lot of hypotheticals in that - but the authors of the paper worked out the details carefully with math and concluded that sometimes everybody is made better off by such a policy. Can we leap from this to the conclusion that Keynesianism is right and that just spending money without regard to where it comes from and who it goes to cures all economic ills? Should we conclude that we should go back to using simple IS-LM models for policy analysis? At least one of the authors evidentally does not think so - the quote is from a paper entitled "The Optimum Quantity of Money Revisited"  and the authors are Tim Kehoe, Mike Woodford and me.

Why doesn't our work vindicate Brad's Keynesian view? For starters we also concluded that "for a broad range of parameter values deflation is clearly good" - that is, sometimes Brad is wrong and sometimes he is right - it depends on what the economy is like. Even when he is right, however, the extent to which everybody is made better off is quite small: the paper shows that for reasonable parameter values if inflation increases by 1% this is equivalent to increasing GDP by .004 percent. Now it might be tempting from this to conclude that the right thing to do is to have a 1000% inflation so as to get a 4% increase in GDP, but that is not correct - in fact for higher rates of inflation GDP goes down rather than up. Worse - if there is too much inflation GDP collapses entirely. The reason for this is worth pointing out: if there is too much inflation nobody wants to save because their savings will be worthless next period and this leads to a coordination failure in which money has no value because nobody wants to hold the stuff - and the economy collapses to no-trade and full unemployment. Notice that this stands Keynesianism on its head: Keynesians think that an economic collapse is caused because people want to save too much and not consume enough - here on the contrary it is caused by people who want to save too little and consume too much. This theory of collapses caused by high inflation has - unfortunately - been extensively tested - for example in the early 1920s in Germany where it led to an output loss comparable to that of the Great Depression.

I do not want to deeply defend a model in a two decades old paper - vastly better models are available now - but it illustrates an important point: if we do carefully the things that Keynesians talk about we do not get simple Keynesian results. We do, however, get results that accord vastly better with the data - indeed one of the great problems with modern models is that often there is not enough data to tell us which one is right. The contrast between this modern literature with Brad's rejoinder to the simple - but precise - example in my essay is striking. He says "the problem in a 'general glut' is that an excess supply of pretty much anything else is the flipside of an excess demand for safe, liquid, reliable financial assets." He then quotes Mill and Malthus speaking of "a superabundance of all commodities relatively to money" and "glutted markets, falling prices, and cotton goods selling at Kamschatka lower than the costs of production." Here is the thing: he speaks of a "glut" as if it is some sort of recognizable object like a human being - but in fact it is a theoretical construct. What does it mean that there are "glutted markets" in Kamschatka? That there are cotton goods lying in the streets unpurchased? Am I to imagine that this "glut" of cotton goods has some independent behavior of its own? I understand people who want to buy and sell and make things now and in the future and who might or might not be rational in various ways. Yet in this discussion of DeLong, Mill and Malthus there is no mention of people and what they might or might not want. Brad speaks of phlogistons and molecules and I find this highly ironic. The 19th century theory of "gluts" sounds to me like "phlogistons" as do the equally mysterious 20th century theoretical concepts of "aggregate supply" and "aggregate demand" - while the 21st century theory of people who buy and sell and produce and trade over time and with uncertain consequences sounds to me like the theory of molecules. Brad apparently feels the opposite.