You will recognise yourself from my description. (Or you will recognise others who fit this description).
You are probably very smart. You are probably very well-educated — either formally, or self-educated, and probably both. You spend a lot of time on the internet reading economics blogs and commenting on those blogs. You maybe even have a blog of your own, where you write about economics topics. You are probably politically engaged. You are probably a lefty, but may be a righty, or someone who is not easily categorised on that political spectrum. You probably think of yourself as a critic of economics, or a critic of what you see as orthodox economics. You are probably sympathetic to what you see as heterodox economics.
But you have never once read a first year economics textbook.
You have probably many times told me, or people like me, that I really really should read something you want me to read.
Well now it's my turn.
I think you really really should read a first year economics textbook.
You have invested so much time in thinking about, reading about, and writing about, economics. Don't you think it would make sense to spend a tiny fraction of that time just reading a standard first year textbook, cover to cover? You could do it in one day. Maybe two, if you go really slowly and carefully. They were designed to be read by an average reader who is probably less smart and less motivated and less knowledgeable than you are. It's gonna be a breeze!
Even if you think you won't agree with it. Even if you think it's going to be all horribly wrong.
Because, at the very least, you will better understand how the people you are criticising view the world.
Because, at the very least, you will better understand the language that is used by the people you are criticising.
Because, at the very least, you will better understand what is and is not an idea that is seen as "heterodox" by the people you are criticising.
Because, at the very least, you would make this blog and other blogs better.
(The final straw that lead me to write this post was reading a comment on another blog which said that "loans create deposits" is a heterodox idea. Every first year textbook I can remember reading contains a description of how loans create deposits.)
This is said with the greatest of respect (and I do mean that, because I really do respect the amount of intelligence and effort some of you are putting in). But also with frustration.
Now, a second plea: does anyone know a free online source for any reasonably good and reasonably standard first year textbook? One that covers much the same topics and in much the same way as any of the best-selling first year university textbooks? Because I know you can pick up old editions of paper copies of the best-sellers for next to nothing, but I also know that people are much less likely to do this.
(P.S. Non-economists may be surprised that I haven't said which textbook I would recommend. That's because it doesn't really matter much. They are all fairly similar in coverage and treatment. And they are almost all good, in my opinion.)
Worthwhile Canadian Initiative 
“The point is that you can still understand how M/B depends on things like the reserve ratio, which is one of the components in m, and that is interesting and informative.”
Yes you can, but no it’s not informative or in any way interesting.
“For example it explains why large recent increases in B have not lead to large increase in M, because reserve ratios have increased.”
That is an excellent example of a just-so story: A purported explanation that obfuscates more than it illuminates. Increasing m is not the reason M hasn’t increased. It’s a consequence of the fact that M hasn’t increased and B has. Pretending that the money multiplier has anything interesting to contribute to thinking about money gives the careless economist a totally back-asswards impression of the causal relationship.
– Jake
from the wiki article on the MM, I read this by Samuelson in 1948
By increasing the volume of their government securities and loans and by lowering Member Bank legal reserve requirements, the Reserve Banks can encourage an increase in the supply of money and bank deposits. They can encourage but, without taking drastic action, they cannot compel. For in the middle of a deep depression just when we want Reserve policy to be most effective, the Member Banks are likely to be timid about buying new investments or making loans. If the Reserve authorities buy government bonds in the open market and thereby swell bank reserves, the banks will not put these funds to work but will simply hold reserves. Result: no 5 for 1, “no nothing,” simply a substitution on the bank’s balance sheet of idle cash for old government bonds.
— (Samuelson 1948, pp. 353–354)
I think Samuelson counts as a mainstream economists, don’t you? Where has this idea come from that economists think the central bank can just raise B and M will rise, regardless?
JakeS: “Rowe: The three-body problem (and, by induction, the n-body problem) is an excellent example of a problem which one can prove has no analytical solution, but which is none the less sufficiently computationally tractable…”
Thanks. Now I understand.
Jake,
I think you’ve gone off track here. Why do you think M has not increased even though B has? I think it’s because banks haven’t increased lending (for reasons I think we all understand, reasons that are the underlying cause why M has not increased). This shows up as high reserve ratios, and once you have taught students the money multiplier you can explain all this. Before you’ve taught it, they don’t know the difference between M or B or why banks lending creates money and means M!=B and so on.
Consider profits=marginsrevenue. If I was to say revenues have increased but profits haven’t, because margins have fallen, that analogous to what I’m saying the money multiplier story is doing here – I agree it’s not an explanation because you still have to explain why margins have fallen. But of course one still has to teach profits=marginsrevenue, it does’t obfuscate or cause anybody to get thing ass backwards. And when accounting relations like these can be broken down into further pieces (cash in hard, deposit ratios etc.) these elements are more helpful. You can say, oh, margins didn’t rise because this line of expenditure here rose. (and you still have to explain why).
Utterly coincidentally, I wrote a post over the weekend, which went up today (Wednesday) on an idea that pops up, such as in historical anthropology, that seems to go back to Marx, that exchange is about “matching equivalences”. Folk who do not get the concept of gains from trade analysing economic behaviour. Sigh
link here NR
(You might like the notion of economics envy.)
Then there are the very strange notions about what mainstream economics covers. David Graeber, for example, claims that non-state bureaucracies is an idea “absolutely foreign to economics”. Strange that Oliver Williamson got a Nobel for studying exactly that.
But if you want a really appalling effort at misrepresentation, try the piece linked below. (For example, did you know that Douglass North got his Nobel in part for his analysis of the origin of money? Strange, neither did the Nobel Committee.)
link here NR
[edited to fix typos NR]
That should be “there are the” and “did you know that”. Clearly, too tired and time to go to bed …
[Fixed. Go to bed Aussie! NR]
“Before you’ve taught [money multiplier], [students] don’t know the difference between M or B or why banks lending creates money and means M!=B and so on.”
But surely you can say something like: “The central bank announces an interest rate. Banks make loans based on that interest rate (and what they expect that rate will be in the future). If the borrower then moves the money he borrowed to a different bank (or demands cash), then the bank goes to the central bank and borrows the needed cash at the interest rate the central bank announced back at the top of the paragraph.”
Then you can add superfluous complications like deposits, the interbank market or reserve requirements later. But this formulation should not mislead students into believing that banks are intermediaries between savers and borrowers. (Though the idea is so widespread in the uneducated public, and sadly even among economists who should know better, that you may want to include a sidebar explicitly stating why it’s nonsense.)
– Jake
JakeS: I was going to suggest you see the Simon Wren Lewis post, where he (at least roughly) agrees with you. But then I saw you had already read it.
But something you said caught my eye: (You were arguing against loanable funds theory) “…there is no causal relationship between your decision to not-buy the TV and your neighbor’s ability to buy the TV.”
In a New Keynesian model, for example, there is indeed a causal relationship. You decide to not-buy the TV, this reduces AD and reduces the central bank’s forecast of inflation, so the inflation-targeting central bank sets a lower interest rate, and so your neighbour decides to buy the TV.
Jake,
sorry to be presumptuous, but I think you seen shortcomings in what you see as the standard story in economics, but you’re throwing the baby out with the bath water. For example I think it’s important to know that banks can lend first and then go to the central bank to fund that loan later, and yes 1st year macro does skip that point (I can’t remember where I first learnt it, but it wasn’t in a macro class). But banks don’t always go to the central bank. Banks go to great expense to attract savers (deposits). Why do they do that? Intermediation between savers and borrowers is part of what banks do.
Incidentally, I sympathize with your accusation on mainly macro that neoclassical economics has wrongly neglected the role of debt. Right, got to get some work done.
DavidN,
I was made aware of Mas-Collel by Keen but got it myself so that I could check.
Luis,
I’m sorry but that section fits my characterisation far better than yours. He does not pause and reflect on how ridiculous it is to assume a dictator. He does not say ‘we must assume something as silly as this, it must be wrong.’ He simply states that it is a necessary condition for the property of aggregate demand we seek to hold, finds a way to fulfill that necessary criteria, then continues.
Very similar arguments can be found elsewhere: in order to satisfy the market demand function, consumer utility must take a very specific form. There are various categories of restrictive assumptions to this. All are unrealistic – e.g. Gorman assuming all consumers are the same – obviously none are as striking as Mas-Colell’s.
But what that formulation misses is that “I not-buy a TV and save the money and he borrows to buy a TV” is not equivalent to “I buy a TV and lend it to him.” Which in turn is not equivalent to “I buy a TV and he decides to not-borrow the money to buy a TV.”
Only the first of those three can contribute to systemic financial instability (unless our loan contract in the second example specifies that he must pay money if he cannot return the TV – in which case the TV can break and we’re back to #1).
Because money isn’t neutral.
The network structure and magnitude of credit relationships matter, the rate of inflation matters to the rate of defaults in the long run, an economy can have a structural rate of inflation, and an inflation-targeting central bank will (unless it also does other things that New Keynesian models don’t even have a language to discuss) not have its eyes on any of these balls.
This is not to say that short-run IS-LM models aren’t useful conceptual shortcuts (the long-run models, OTOH, try to force-fit a complex dynamic system into an equilibrium straitjacket). But the reasoning used to derive them is full of epicycles. Presenting them as mnemonic devices backed by black empiricism would be far more useful than trying to put them on Walrasian microfoundations.
– Jake
JakeS: “Presenting them as mnemonic devices backed by black empiricism would be far more useful than trying to put them on Walrasian microfoundations.”
I tend to agree with that. Putting a model of a monetary exchange economy on Walrasian foundations is, in any case, let’s say “problematic”. If we had a single centralised market with a Walrasian auctioneer, we wouldn’t be using money. (Clower vs Patinkin). (Or we could think about non-Walrasian microfoundations, even though it’s harder.)
Also, Nick, I forgot: a day to read a textbook? Are you Matt Damon from Good Will Hunting?
“But banks don’t always go to the central bank.”
Comes to the same thing in the end. If you’re going to start simple, start with the counterfactual where there are not deposits. It is no more counterfactual than the money multiplier story, and is less likely to lead people up the primrose path.
“Banks go to great expense to attract savers (deposits). Why do they do that?”
Because they make money on the spread between what they pay you and the CB policy rate. And on shipping and handling fees: Just yesterday, I paid about € 3 to change € 66 worth of DKK into SEK, and I pay something like that every time I use another bank’s ATM instead of my own bank’s. Both of which are very nearly pure profit for my bank.
And because regulators like to use loan/deposit ratios as a shorthand for identifying banks who have an unbalanced growth model. Something that has historically been indicative of a Ponzi bank. And occasionally just for the sake of chasing deposits, because that’s what banks do because they do it. There was a hilarious episode in Spain not so long ago where banks were, for weeks and months together, paying depositors more than the Spanish central bank’s penalty rate – that is, they were losing money on every € of deposits. And they were still trying to capture market share.
For historical reasons, the deposit-taking business cohabitates with the lending business. But with a modern central bank there’s no reason in principle why this has to be so. Banks like that arrangement, however, because it offers a partial hedge against wrong forecasts of the CB policy rate (your deposit business will lose money, but your lending business make money, or vice versa). And because it makes people afraid that they may not be able to get their paycheck if the financial regulator takes a bank out back and shoots it.
– Jake
Unlearning: yeh, maybe I was a little overoptimistic there. But Daniel Kuehn’s story (very short), and one of his other commenter’s, is so so similar to my own. All 3 of us did it in basically one sitting. And it made all the difference.
I know others here are giving you helpful advice on Varian etc. But somehow,….I disagree, and feel…it’s just not the same as that First Year Textbook In One Big Read Experience! Where it all comes together, and you get the Big Picture, even if you query all the details. And everything else, up to Mas-whoever, is just a long footnote plus math appendix. (Slightly unfair to upper year texts, of course.)
Unlearning
so what do you think students take away from that passage? What would you say about it, if you were teaching that class? That’s it’s A-okay to treat aggregate consumer surplus as a measure of welfare because we can just assume there’s a benevolent social planner redistributing before trade and forget about distributional questions?
you presume economists are idiots then find that they are idiots.
Luis,
I think at this point students are so convinced that assumptions don’t matter that it is just glossed over. I don’t think economists are stupid AT ALL. Was it George Orwell who said ‘that is so ridiculous only a smart person could believe it?’
What is your experience of this in class, anyway?
LOL, when mandos asked on the „kill the euro“ session, “where is genauer?“ I kept my keyboard shut, because that discussion became repetitive and people apparently do not learn anything from each other.
0. “read an intro textbook first”
made me first sympathize a lot. I had these situations, when people come, and want to discuss with me about gravity, and “that we do not really understand it”. You check with them, that they do not even know the 10th grade 1/r^2 law of force, that is actually not on the curriculum for about 75 % of the people, and if you test people over the age of 40, my guess would be that only about 5 % of the population can still recall that by themselves. I tell them that for all practical purpose, we understand gravity perfectly, and there is nothing to discuss. But that I can explain them how good we know it, and that it might be useful, that they read a (quantitative) intro textbook first, and not some “popular science” musings of Stephen Hawkings about string theory and the likes. Then these folks find me arrogant, that I don’t accept them as equal discussion partners, and even doubt my knowledge, despite me telling them about my physics PhD. Pretty interesting experience.
The same folks think that political correctness and majority votes trump scientific truth, that if 10 people including one “forest scientist” tell me that tropical rain forest is a net absorber of CO2, because Greenpeace says so, that I have to accept this, and can not even understand the argument, that “mass conservation” dictates that forest without any significant humus underneath can not be a significant CO2 absorber in the long run.
It would probably be too painful for them, to understand, how severely their stupidity and ignorance limits their intellectual horizon.
1. N-body problem
Now here, the n-body problem illustrates a lot of beautiful things in physics.
Formulas and laws are exact, and there is a very clear direction between cause and effect. The (gravitational) force on a body is a function of its mass, and not just one correlated to the other, in strong contrast to e.g. economics.
That also leads to a lot of exact “conservation laws” mass, energy, impulse, spin. And from those you can derive, that as long as you only consider the 2-body problem, like the earth around the sun, the earth will never ever leave its orbit around the sun, and disappear in the cold, dark rest of the universe. Unfortunately, the same can not be said with the same exact, universal rigidity, we physicists love, as soon as you add a 3rd or more bodies. These systems have, just in principle, the possibility to fly apart, let’s say Venus and mars descent into sun, and transfer their momentum onto the earth, which then exits the orbit, well, yeah, in principle, theoretically, this can not totally be excluded, to happen, far into the future.
But for all practical purposes you can treat the problem of 8 or 9 planets as “weakly coupled” 2-body problems, where the other planets are just weak perturbations, and you can come up with pretty precise predictions with very little calculation effort. And this is very typical for physics and engineering.
“k”, when you refer to some numerical simulation toolkits, you simply don’t understand the problem. If you would have ever tried to calculate / simulate even such a simple thing like only the earth orbiting the sun, in x-y-z-t coordinates, you would be very surprised how fast the numerical errors add up, and produce strange trajectories.
This kind of solving a problem is something “economics” can not even dream of, since none of the “laws” are anywhere near “exact”, not even the accounting “identities”.
2. The Sokal hoax and related things
What Sokal demonstrated, is that in a lot of Social science studies “political correctness” and the right “progressive” attitude replaces any meaningful content. Especially since they could not just be ashamed of it and keep their mouth shut. Hard science people assumed that all along, but this was final proof.
But in the meanwhile we also had a couple of incidents, where people in real science were found cheating, over interpreted data far beyond reason, and also engaged in meaningless ramblings, just dropping the right words, concepts, without adding anything meaningful (“Bogdanov affair”). One guy (“El Naschie”) even managed to run a full Elsevier Journal for a while. Apparently he is still running his show in his homeland Egypt. A lot of (more experimentally oriented) physicists say, this is the typically result of “research fields” where basically a (near) complete lack of experimental results like in elementary particles is replaces by ever wilder theories (string theory, m-theory), where even 30 years later nobody has any idea how to prove/ disprove anything and what some people like Woit characterize as “not even wrong”. But combine it with the “intelligent cripple” meme, leave out any equation and real data, and you can sell “popular science” books like fresh bread.
Now, do you see the close parallels to what Paul Krugman does? Carefully avoiding any contact / measure to the real world in his papers? Based on what real world fact could you falsify any of them? You can’t. Ever heard of Sir Popper?
3. Lots of equations & a big computer = understanding a problem?
Whenever I see this as a physicist / engineer I get a big laugh out of it, and then I feel sorry for these people. First you understand the underlying mechanisms exactly, and then you can put it together in larger simulations of finite elements, but only if there is a need because of weird geometries, or really complex dynamics. In the very most real world cases, you have a very few dominant factors, which give you good first order approximations of “the solution”.
It was one of the few important impacts of Hayek to rant against the folly of this soviet “central planning” while it was all the rage with the economists of that time.
And while we are at that, people calculated planet and star trajectories, used vapor pressure / temperature curves and equations to built steam machines long before they knew the “microfoundations” Newtons laws or even knew about atoms and their mechanics.
All those “philosophers of science” who later on want to define how real science progresses, are pretty hilarious.
The current German pope actually did read Paul Feyerabend “Against Method: Outline of an Anarchistic Theory of Knowledge”, who I warmly recommend as an antidote against those orthodox charlatans.
Sooo, I was pretty open to how things work in economics, read the text book Felderer-Homburg for macro and Gärtner-Lutz for exchange rates (German books, and well, the level more like US PhD) I was pretty disaapointed by the 3 or 4 Krugman books, pretty expensive, very low level, 80 % of the content was the same, and very rarely any real world data. One of the rare cases, where I didn’t buy and just borrowed them from the library.
More is following in maybe 2 hours, or tomorrow.
genauer, whether you can treat planetary motion in the solar system as eight weakly coupled two-body problems or you need to drag out the big chaos theory guns depends very much on what time scale you are operating on. Dragging out the Poincare plots is silly if you’re planning a space mission. Not so much if you’re modeling on an astronomic time scale.
Likewise, if you’re modeling the current Eurozone crisis, all you really need to pay attention to is the demand deficiency created by German wage dumping, and how it interacts with a currency peg. You can pretty much disregard all issues internal to the victim countries, because their inability to dump wages faster than Germany dominates any internal merits or dysfunctions they might happen to have. But if you’re trying to formulate a sustainable growth strategy for Southern Europe, not just proving that the Bundesbank is staffed by crazy people, then you need to look at those countries in detail.
Tl;dr: Make it as simple as possible, but no simpler.
– Jake
@Luis Enrique
Nan means “not a number”. It does not mean “variable”.
what I forgot about the n-body problem, the historical context.
I dont see this mentioned anywhere, but I think this was driving this significantly.
As long as you just see the Sun going up every day and down, and assume some godly force behind that, you don’t have to worry about this going on forever.
In the moment you understand that this is just some coupled differential equations, you get the problem: is this stable, or do we circle off in like 12 years from now ? Something most of us want a very clear and assuring answer to : -)
Like today: past results of stocks don’t predict (necessarily) the future.
And some radically exact answer “this is never ever going to happen” is much, much more comforting than: “it is not going to happen at least the next 200 years” although nobody of us will live in 200 years.
Compare this a little bit to calculations people made,
a) before blowing up the first nuke, making sure, that not the whole atmosphere of the earth goes with it
b) the ozone hole, making sure that we dont turn the earth into some irreversible cancer zone, for us humans and other animals, veggies. We phased out these gases pretty quickly, in contrast to the CO2 / global warming situation.
c) before they started the LHC, they also checked that we dont create a black hole, turning the earth into mush instantly
Nick Rowe: “Regardless of this, the market economy can be thought of as “trying” to “solve” some “problem”, and that problem is likely to be at least as complicated as the central planner’s problem.”
The same can be said of life. 🙂
Yes Min, that was the point
genauer: “This kind of solving a problem is something “economics” can not even dream of, since none of the “laws” are anywhere near “exact”, not even the accounting “identities”.”
I remember reading (I think) an Isaac Asimov story once. (Nightfall?)It was about a planet which had 7(?) suns. The 7-body problem. Which made astronomy very complicated, and slowed down the development of Newtonian mechanics. Maybe economics is like that planet. Sometimes God gives you solvable problems; and sometimes he doesn’t.
“Lots of equations & a big computer = understanding a problem?”
That worries me too. I really sympathise with the research program of “agent-based modelling”. But does it really count as understanding (I would use the word “verstehen”, and not just because you are German) the problem?
In case anybody cares, I read a textbook:
[link embedded here NR]
Nick,
Isaak Asimov was nearly 300 years after Isaac Newton. He was just some nice to read science fiction. I dont know about his historic books. That makes it pretty safe to say, that he had certainly zero impact on any scientific thought in astronomy. Reading the wiki of Newton is actually pretty interesting. Practically all these folks like Newton, Keynes, were also involved in other real world things, stock market, hanging counter feiters, unlike most nowadays academics. I see this actually as a real problem of present day academics, too much seclusion.
Test
more econo-mist
If you want to understand how present day physics (as a profession)works, first reading intro and then advanced text books is certainly the necessary beginning. You pick up the basic vocabulary, the basic concepts, get an overview. But most people get then the idea, just because the text books show all the things as completely settled, exact and undisputable, derived from eternal logic, simple experiments, that this would be the way research works.
LOL.
In practice, things work significantly different, often, at the frontier, things are vague, ambiguous, there are more than one way to explain things. You learn a lot more if you read old and new papers, visit conferences, are embedded or close to active research.
Soo, getting into economics, I did the same as I did 20 years ago during my physics PhD, read a lot of old and new papers (about 400), try to make sense of them, implement crucial things in own calculations, create or dig up data, compare models to data, and think about it. Engage in online (blog) discussions of actual problems, ask questions, put out statements and see what kind of reactions you get.
The blog discussions showed, that, not really surprisingly, the open discussion is dominated by the “progressive” people, like Krugman, deLong. The “conservatives” like Solow, Barro, Mankiw prefer discussion, where I do not really have access to, but were the real discussion / decisions are made. OK.
Just a year ago, I also assumed that the guys at the OECD and IMF do know what they are doing. Silly me.
But where things really turned ugly for me, was,
a) when I took a closer look at “crucial”, “seminal” papers. NBER, AER, Brookings, Cobb-Douglas, Solow, Kaldor. And their reception by Krugman, Barro, and so many others.
b) Looking at the descriptions of central bank models like FRB/US and the ECB /SDGE
I went into this not with an attitude of “let me just find something to bicker about”, but just the opposite, “let me learn, how this is done by the very experts”.
And I found a lot of shit, sorry. Not just sloppy thinking, little arithmetic errors, but fundamental flaws, lots of them.
I will elaborate on this tomorrow. Would be nice if some folks are interested in a detailed discussion, why I believe that stuff is WRONG.
But for tonight, for all the folks who think that directing their wrath at the Buba, Merkel, Schäuble does them any good, let it be known, that it is people like me, who looked at the anglo economic thinking, know a few things about capital markets and history. It is people like me, who are working to put more steel into the spine of German officials, to let things rock, if not avoidable. Ordnungspolitik will prevail.
““k”, when you refer to some numerical simulation toolkits, you simply don’t understand the problem. If you would have ever tried to calculate / simulate even such a simple thing like only the earth orbiting the sun, in x-y-z-t coordinates, you would be very surprised how fast the numerical errors add up, and produce strange trajectories. ”
“genauer,” when you assume that somebody doesn’t know what they are talking about because of your misunderstanding based on a cursory reading of what they said, you come across as a presumptuous, pompous ass.
In fact, lots of those toolkits contain excellent integrators for limiting numerical error, even for articulated rigid body dynamics with multiple surface contacts which are quite complex and numerically challenging problems. But why are you talking about numerical stability, when the issue was algorithmic complexity of the n-body problem? Yes, every applied mathematician knows that even simple ODEs will blow up when confronted with a crappy (eg Euler) integrator. That has nothing to do with the difference between 2 and 3 bodies. Even a point mass on a Hooke’s law spring will do that.
Anyways, please don’t answer. I have literally zero interest in debating this or anything else with you.
k,
please contact a good physicist in your vicinity. She can explain to you, why I can make such a categorical, hard statement, that no numerical simulator, you referenced to, can ever deliver a fundamental “solution” to the n-body problem, the people sought for, no matter how sophisticated it is.
You could learn, how powerful physics is, in certain areas. It may not be up to discussion by you, you either learn it, or keep to your assumptions.
And to just finish this, insults do not get you anywhere with Germans, we are very much used to this. Wanna buy a BMW ? We have a special price for people as sophisticated as you are.
K,
Yes, I understand that the computation is distributed across the economy, but there is no way that each firm is actually carrying out their bit of that exact computation, somehow computing the correct price of their good as a function of all the other goods in the economy. In reality each firm makes very rough heuristic decisions based on the price of their inputs and those of their competitors. Wash and repeat.
In an ironic twist, the capitalist system solves the production problem by being wasteful, or at least less efficient – by usually making too much of everything, including a bunch of stuff nobody wants. Sure, companies are punished for making stuff nobody wants, but significant real costs were incurred anyway – see “Aztek, Pontiac”. Some elements of competition may be efficient, but it solves the problem of what people want and how much they want of it in a brute force manner, throwing a boat load of production against the wall and seeing what sticks. Way less efficient than an imaginable omniscient central planner.
JakeS,
You get a much richer model, as well as one more grounded in the reality of things, by dispensing with optimization and replacing it with behavioral heuristics. At no loss of computational power. (But at great cost in survey effort and econometrician-hours to actually observe and codify the common behavioral responses to stimuli.)
Agreed. Behavior is where it’s at.
Nick Rowe,
Whether or not the market economy is actually solving the problem we would want it to solve (The First Theorem of Welfare Economics) is a separate question. Regardless of this, the market economy can be thought of as “trying” to “solve” some “problem”, and that problem is likely to be at least as complicated as the central planner’s problem.
Yeah! A highly idealized “economy” with a realistic number of products and the complication of geography turns out to be way, way too big to tackle. The real economy of market countries, with vastly more complicated and decentralized production and investment decisions, with the terribly bizarre effects of money and money-like substitutes, and with everything responding to the behavioral ticks of the (editorializing) incredibly capricious public? It’s even crazier and less mathy.
Which isn’t to say I don’t think there’s value to realistic elements in the models. Just like it’s self-evidently important to pin down the weirdness money introduces into economic interactions, you’re probably not going to get good results if you’re not trying to capture the institutional framework of the economy and how that affects economic reality.
All analogies are deeply suspect, but it sort of seems like current economics is akin to a heart surgeon who responds to a question with, “Well, we have a lot of models to represent what’s going on in your heart.” We don’t need to know where every blood cell is going at any given time, but it would be difficult to make accurate predictions without taking a serious look at the physical structure the heart. I tend to think that’s where Friedman’s methodological point of view is on shaky ground. Although one’s incredibly unrealistic model may have made accurate “predictions” of past events, I can’t see how concessions to reality would negatively impact its accuracy in the future as the economy is buffeted by new and different shocks.
Okay, I can conceive of ways it could happen, but I think realism as a goal is probably good.
genauer,
“no numerical simulator, you referenced to, can ever deliver a fundamental “solution” to the n-body problem, the people sought for, no matter how sophisticated it is.”
“fundamental” is not a mathematically well defined term, but if you mean that the dynamics are chaotic (therefore solution converges only pointwise), then yes, that’s what I said above. The funny thing is, I really do know the numerical techniques relating to classical mechanics, and have spent a great deal of time implementing them (and yes, I have degrees in physics, but I do not see that as a prerequisite for someone to have a strong understanding of it – people learn all kinds of stuff by themselves). The reason I brought up the dynamics toolkits was because we were discussing algorithmic complexity, and it seems to me that the existence of lots of dynamics toolkits that scale extremely well for large numbers of bodies, is ample evidence that the difficulties of developing 3-body solvers has nothing to do with algorithmic complexity (and it doesn’t for theoretical reasons that we could get into if there was any point).
Also, (and I’m repeating myself here), the difficulty of the 3-body problem is not a numerical stability problem, either. The dynamics are inherently unstable in the sense that any initial misspecification of the state, no matter how tiny, will be amplified exponentially by the dynamic. So, no matter to what accuracy you know the initial state, it will in general be a very short time before you know nothing about the state of the system (other than constraints implied by the conservation laws). Even if you have God’s integrator.
Numerical stability is exactly as important to the 2-body problem as it is to the 3-body problem. If you have a decent integrator (anything better than 2nd order like Runge-Kutta with an adaptive step size will do), then so long as the error due to your integration steps is well within the bounds of the given uncertainty of your problem, then you will obtain a perfectly valid solution. In the 2-body problem, that solution happens to be Kepler’s laws. In the 3-body problem, it’s one of any number of very different solutions consistent with the uncertainty of the initial system state.
But note that none of the difficulties of the 3-body problem are computational in nature. They are intrinsically related to the inherent uncertainty of the initial state specification and the chaotic nature of the ODEs. It’s a physics problem, not an applied math problem. Which is why I suggested to Nick that his analogy was imperfect.
But I get it. You are smart, and you know something about numerical problems, and you want to tell us all about it. But save it for when what you know is substantial and relevant to the conversation. And make a clear point. Blurting out whatever you happen know that seems to appear superficially relevant to the topic for no purpose other than asserting your intellectual superiority to those who don’t know better makes you look like a presumptuous, pompous ass. That’s not an insult. It’s helpful advice.
K: “But note that none of the difficulties of the 3-body problem are computational in nature. They are intrinsically related to the inherent uncertainty of the initial state specification and the chaotic nature of the ODEs. It’s a physics problem, not an applied math problem. Which is why I suggested to Nick that his analogy was imperfect.”
Ah! I think I understood that. So if you give a tiny nudge to a 2-body system it won’t make a big difference to where it will be next year. But if you give the same tiny nudge to a 3-body system it could make a massive difference to where it will be next year?
(But when the n in the n-body problem gets very large, so you are looking at the Ideal Gas Laws, it all becomes simple again, but in a different sense, because the averages are simple but the individual molecules are impossibly complex?)
When you look at the ideal gas law, you’re not looking at the position and momentum of individual atoms and molecules. That would be impossible (and not very interesting). What you are looking at is the statistical distribution of energy (and, in the grand canonical ensemble, of particle number), given the boundary condition of temperature (resp. temperature and chemical potential).
While one should be careful with analogies between physics and economics, temperature is a bit like interest rate, while money supply is a bit like internal energy. And unemployment is like interest rate and chemical potential, while the sovereign deficit is like money supply and particle number: If you set a target unemployment and let fiscal policy adjust, then you will hit that unemployment target. Regardless of any determinant of individual firm behavior, such as “the state of confidence.” Whereas if you attempt to target individual firm behavior, you have to deal with the incredibly complex system of firm interaction – a much more difficult problem to solve.
I’m not sure inflation has a simple analogy, because it’s not a state function.
(In the limit, of course, targeting macroeconomic aggregates can prompts a coup d’etat by either entrenched rentier interests or the parliament of the street. All models have domains of validity; if you try to increase internal energy while keeping particle number constant without regard to the design tolerances of your pressure vessel, it will eventually explode and ruin your attempt to impose the desired boundary conditions.)
And if you’re thinking right now “let’s go read a stat-mech book and do awesome economics” then fuggetaboutit. Stat-mech requires a very detailed understanding of individual particle behavior, and we do not understand firm or household behavior to such a level of nicety (and may never do). What we can do in macro, given the state of psychology and microeconomics, is describe the relationships between aggregates, sort of like (non-equilibrium) classical thermodynamics does. And those can be plenty complicated enough to keep things interesting for several lifetimes.
– Jake
JakeS: My hope is that there just might be something vaguely like the Ideal Gas Laws for macro (even if nowhere near as good). Where we can say little or nothing useful about any individual, but we might be able to talk about probability distributions and averages. We might be looking for our keys under the lamppost, but we don’t know in advance they are up the other end of the street. And if they are, we are screwed anyway.
“And if you’re thinking right now “let’s go read a stat-mech book and do awesome economics” then fuggetaboutit.”
I sure wasn’t! But sometimes analogies help.
k,
your initial
“Nick: “physics can solve the “2 body problem” (Newtonian mechanics) but has runs into problems for numbers bigger than 2″
I wouldn’t say that”
and your continuing “If you have a decent integrator … then so long as the error due to your integration steps is … then you will obtain a perfectly valid solution. In the 2-body problem, that solution happens to be Kepler’s laws”
That shows that you still dont understand the fundamental difference between a numerical calculation and an exact, analytical solution. Since you seem to believe to understand something about physics, apparently a professor for theorectical physics is needed to teach you the difference. I am very aware that people like you dont like that kind of advice, and that is is totally hopeless that you ever accept the explanation from me, after you started with your insults.
Cool it. Don’t show me that physicists are even worse than economists!
Nick,
I think different people have different understandings of “understanding” (“verstehen”). For myself, a quarter century ago, it was also simple: just calculate the exactly one right answer to all (math /physics) exam questions. Not just talking. Do you remember the Turing Test / the ELIZA program ?
For me being able to calculate a solution depended on having understood the underlying math /physics. Apparently, especially with the multiple choice tests today and strict bodies of problems presented, there is a kind of folks, who are perfect at getting the right answers to those standards, but fail to do any transfer, (e.g. http://ftalphaville.ft.com/blog/2012/07/26/1096691/top-quant-to-next-generation-you-suck/) which is practically the main thing you need in real life, eeeerm, at least as part of the “creative class”. And also to understand that there might be “no solution” or “we dont understand this at all”, yet or forever. And to not replace these answers with some general wishy-washy or guesses.
Soo, clearly, about agent based modeling, by far, I dont know enough to utter any kind of opinion.
Isn’t the example of spades and men that unlearning gives just an example of the leontief production function? Which as far as I can see is just a variant of the neoclassical production function.
http://en.wikipedia.org/wiki/Leontief_production_function
I like Jake’s comparison of temperature etc to interest rates to a certain degree. But I would like to add: electrons and atoms behave exactly the same, each one (of the same type), now and in the future.
People have feelings, talk to each other, and sometimes actually learn from the past. Not necessarily the right things. And that makes economics a lot more exciting / difficult.
Maybe we should formulate some kind of pseudo conservation law:
Somehow the finance sector always comes up with some thing new, to screw up at least a region every 10 to maximum 20 years.
The analogy between interest rate and temperature should not be taken overly seriously, obviously (that was my point with the “don’t read stat-mech and think you can do econ” line). But it illustrates well the differences between intensive and extensive variables and how you can enforce boundary conditions without ever touching the micro behavior directly.
(On that note, I made a cock-up above: The analog to chemical potential is government debt, not government deficit – the analogy to gov’t deficit is flow rate across the boundary.)
The law you propose already exists: “Stability is destabilizing” was Minsky’s wording of it. Or, as a friend of mine once put it somewhat more colorfully: “We repeat the mistakes of our fathers, the gross blunders of our grandfathers and the catastrophes of our great grandfathers.”
This, by the way, should not be at all reassuring for those of us who are aware of the parallels between Brünning’s defense of the gold standard and what Merkel, Weber and Stasi 2.0 are doing right now.
– Jake
I understand your dislike of Merkel.
But :
Weber, Stasi 2.0 ?????
Huugh, who, and what has this to do with what?
With regard to Brüning and other stuff 80 years ago. This is perceived in Germany as scare tactics, and pretty much worn off.
genauer: I currently writing some sort of post on verstehen and agent-based modelling.
Gavin: I agree. It’s a limiting case. At one extreme you have perfect substitutibility between inputs, and at the other extreme perfect complementarity.
Jake: “We repeat the mistakes of our fathers, the gross blunders of our grandfathers and the catastrophes of our great grandfathers.”
Beautiful!
With regard to Brüning and other stuff 80 years ago. This is perceived in Germany as scare tactics, and pretty much worn off.
And yet, being unreasonably terrified of an unlikely inflation is not a scare tactic, but the conventional wisdom.
Nick,
I look forward to the agent-based modelling post.
I remember sitting down with my PhD supervisor and explaining to him that I wanted to do agent-based modelling. He explained to me that unfortunately the attitude in the profession tends to be: nice computer game, why I am supposed to be more interested in it than in Civilisation 5? [actually probably Civ II back then]
now of course that’s a dumb response (he was parodying) and there are of justifying your choices in agent-based modelling just like in any other approach, and there are plenty of very interesting agent-based models out there already to build on, but I was advised there is a real problem here, which is that essentially few economists really know quite what to make of it all, which makes it a poor choice for young researchers trying to land a job.
I hope to move into agent based modelling once I have a secure job.
“Isn’t the example of spades and men that unlearning gives just an example of the leontief production function?”
No, Dennis Robertson’s example is not of a Leontief production function.
Anonymous: A Leontief production function can be written as Q=min{aL,bK}. (Equivalently, Q=min{L/a,K/b}
Let a=b, and L be men and K be spades. One man+spade combo digs a square metres of land. An extra man with no spade digs nothing, and an extra spade without a man digs nothing. Did Dennis Robertson say something different?
“The Leontief production function as a limiting case of the CES”
“Weber, Stasi 2.0 ?????”
Weber, Axel: The last president of the Bundesbank, and former member of the ECB’s politburo.
Schäuble, Wolfgang (a.k.a. Stasi 2.0): Current minister of finance, former minister of internal security for the Bundesrepublik (where he earned his nickname for his suggestion that the political police should take scent samples of demonstrators, just like Stasi did in its time).
“With regard to Brüning and other stuff 80 years ago. This is perceived in Germany as scare tactics, and pretty much worn off.”
And thus we repeat the catastrophes of our great-grandfathers.
Except, of course, this time it’s Greece that’s getting the Versailles treatment.
– Jake
My favorite is how tax cuts lead to more investment and growth because they lower the deficit. How wait – that wasn’t a comment at some economics blog. That was Glenn Hubbard in today’s Wall Street Journal. Of course his textbook would not have been so misleading. So buy Glenn’s textbook and just skip his WSJ oped!