The Lucasian map is not the Hayekian territory

In defence of Lucas '72.

Take any macroeconomic model of a market economy with inefficient aggregate fluctuations. In fact, take any economic model where something bad might happen.

Assume that model is literally true.

The people in that model are idiots.

This conclusion follows immediately. If they weren't idiots, the people in the model would appoint the economist modelling the economy as central planner, who would tell them all what to do, and make them all better off.

The people in Lucas' '72 model are complete idiots for producing less because they don't realise there's a recession on.

The people in New Keynesian models are complete idiots for waiting for the Calvo fairy to give them permission to cut prices in a recession.

All models suffer this same problem. If the world really were as simple as the economic model of that world, people would figure it out, and wouldn't let bad things happen.

It's unfair to single out Lucas '72 for this criticism. Actually, it's doubly unfair, because Lucas did at least address this question, and carefully rigged up the assumptions of his model so that the people in it wouldn't be able to figure out that there's a recession on. Was that a satisfactory answer? Not really, because most people do read the news, and know there's a recession on. But at least Lucas gave an answer.

The map is not the territory. All models are simplifications of reality. They leave masses of stuff out. That's what makes them models. That's what makes them (potentially) useful. But it's also what makes any model of a market economy self-contradictory. If the economy really were that simple, people wouldn't need markets to resolve the Hayekian problem of the coordination of the changing plans of multiple people, each with their own local knowledge.

All real world markets, even the simplest, are more complicated than any economic model of a market.

Take the housing market for example. If I wanted to model it, especially in a macroeconomic model where all complications get squared or cubed when we add them all together, I would use something like a simple demand and supply model. But at the same time I would know that's hopelessly oversimplified.

Suppose you decide to sell your house. You know that your house is different from other houses on hundreds of dimensions that might matter to a potential buyer. You also know that each potential buyer is different, and evaluates each of those dimensions differently from any other buyer. When you post a price, and when you decide whether to accept an offer, or make a counteroffer, you know you face a trade-off between getting a quick sale and waiting for a better price. And each potential buyer faces a trade-off between buying your house now or waiting for a better price or a house that suits him better. And the trade-off facing one seller, and the trade-off facing one buyer, depend on the choices that will be made in future by all the other potential buyers and sellers, which depend in turn on their perceived trade-offs.

That's an incredibly complex market. And I'm sure I've left a load of stuff out.

The labour market is even more complex than the housing market. All that matters to the person selling a house is the price and when he sells it. (OK, that's an oversimplification). The person selling his labour has to think about where he will be spending half his waking hours for the next few years, or decades. And I'm sure I've left a load of other differences out too.

Start in equilibrium (whatever that means in a housing market like I've just described). Now imagine that there's a tightening of monetary policy. What happens?

"God only knows" is the short answer.

Look, monetary economists can't even agree on what "monetary tightening" means. How do you expect the average home seller or buyer to understand it, even if they do all read the newspapers?

In a very simple model of the housing market, and if all markets were equally simple, and if everybody understood monetary policy, I know what would happen. The demand curve would shift left; the supply curve would shift right; and the nominal price of houses would drop in proportion to the drop in the money supply.

At least, that's what I think would happen, but I bet some people reading this will already disagree.

And I'm fairly sure, but not 100% sure, that something vaguely similar to this would happen eventually. After all, for any equilibrium there ought to be another equilibrium where all the monetary units are changed but all the things measured in real units are the same. I'm not 100% sure because there might be more than one real equilibrium.

But in a real world housing market, like the one I have sketched above only more complex, where most people don't understand what's going on, and their mistakes will affect the trade-offs facing other buyers and sellers, and so affect their behaviour, which again affects those trade-offs, and leads to more mistakes.....? I don't know what's precisely going to happen. Nor does any other economist. And, more importantly, the buyers and sellers in that housing market, and all the other markets that interact with that market, won't have a clue what's really happening and what their optimal response should be.

The signal-processing problem facing real buyers and sellers in a real market are massively more complex than the signal-processing problem faced by agents in Lucas' model in a perfectly competitive market of a homogeneous good.

"It's taking me longer to sell my house than I thought it would. Am I pricing too high? Did I just get unlucky, and the person who really wants to buy my house and is prepared to pay my price just happens to be coming a little later than normal by sheer chance? Has my trade-off between getting a quick sale and getting a high price gotten flatter or steeper?"

And potential buyers are asking similar questions.

And no economist knows the answer to those questions, so you can't expect the buyers and sellers to.

That's even before you throw in something like menu costs of changing prices.

Suppose recessions lasted 1 week. If they did, I think most economists would have a very different view of Lucas '72. "It takes people 1 week to figure out there's a recession on and how to react? Sounds plausible."

Suppose recessions last 100 weeks. "It takes people 100 weeks to figure out there's a recession on and how to react? Sounds totally implausible."

But maybe real world markets are 100 times more complex than the simple demand and supply model. And maybe the signal-processing problem is 100 times more complex than in that simple model. And maybe it takes people 100 times longer to figure out how to react, when one person's reaction depends on everyone else's reaction. That sounds plausible to me.

We draw a supply and demand curve and point to where the two curves cross. We shift the supply and demand curves and point to where the two new curves cross. The only people who talk about the process of getting from the first point to the second point are: teachers of Economics 1000; Austrian economists.

Maybe it takes time to get from the first point to the second point. Not because people are idiots, in not changing their prices, or not figuring out what's going on and how to react. But because the territory is a lot more complicated than the map.

And it's always going to be hard to build a map of how the territory is more complicated than the map.

Comments

[Frances Woolley]

Nick: "The people in that model are idiots." ;-)

I wish you'd been blogging back when I was taking grad macro, and Lucas et al were all the rage.

[Nick Rowe]

Lucas sounded good to me at the time. Then, when the 1982 recession hit, it didn't sound so good. I became a New Keynesian. But I've always wondered. Maybe Lucas had the right story, but we just aren't telling it right? And if we really did tell it right, and if we really understood the sticky price story properly, maybe the two stories would end up being the same?

[Nathan Tankus]

this is what happens when economics drowns in the vacuum of it's own thought. "All models suffer this same problem. If the world really were as simple as the economic model of that world, people would figure it out, and wouldn't let bad things happen."

you are implicitly assuming that in all economics models, all agents have complete and equal control over the result of the model. this is only true when there is no power inequality among agents. you are assuming your own conclusion. if i made an economic model of slavery however, the slave agents in that model would almost certainly be in a very suboptimal position while the slave-owning agents would be the agents able to reach an optimal outcome (assuming this model abstracted from agents with power over slave owners).

obviously I'm not saying that the power relations between slave and master are akin to the power inequalities we are experiencing now, but there are power inequalities that affect economic relations and just because you abstract from them doesn't mean an economic model can't include them.

[TGGP]

Nathan, who do you think is better off when a recession occurs?

[Nick Rowe]

Nathan: Most of these models are 'representative agent" models anyway, so all agents are equally affected. Even if they weren't, an "inefficient" outcome means (roughly) it's possible to find a deal that would make all agents better off. The winners could compensate the losers, under the deal.

[Nathan Tankus]

@nick i know this. my point was that the power equality assumption is one of the many assumptions underlying these types of models and only under that assumption (well to be fair, the assumption of a future reducible to probability distributions known by the agent is probably also needed) could agents be considered "idiots" for not reaching an optimal outcome.

@tggp .not necessarily anyone. at least i haven't done the kind of political-economic modeling I'm interested in yet to come to an adequate conclusion. I do think however, that that is the wrong question. an importer might hope for a recovery in their customer's incomes, but he/she would certainly oppose import quotas aimed at that goal. this is just an example. i obviously didn't use that example to exclaim that it is importer'ss fault.the extensive archiving and research in a book like this one:http://www.amazon.com/Golden-Rule-Investment-Competition-Money-Driven/dp/0226243176 however, seems promising (it is a very good read and i suggest you process the argument even if you reject the conclusions).

[pe]

Angeletos and La'O are kind of trying to grapple with these issues dont you think? Do you like their
attempts?

http://econ-www.mit.edu/files/7013

[Kevin Donoghue]

I think a large part of the problem with macro is that economists have adopted a ludicrous set of rules regarding the way the 'maps' are drawn. Lucas must bear a great deal of the blame for this. It's as if all map-makers were constrained to use a maximum of two colours or something daft like that.

One reason why recessions drag out is that we can always see more than one way in which they will end. For example, the Eurozone can disintegrate, or survive with a new set of rules. Whether it is a better idea to locate a new factory in Lyon or in Turin may depend on which outcome transpires. So it just doesn’t get built until the situation is clarified. If in doubt hold cash.

[Patrick]

Is 'understanding what's going' even possible? I suppose that's kinda what you're getting at.

'What's going on' is what people actually do and what they expect will happen and how they plan to react to what people might do, maybe. But they aren't sure. And they might do something else, for totally unknown reasons when the time comes to actually do something. And I end-up sounding like a passage from Winnie the Pooh.

My 'doing' is only visible to a few people around me, and my expectations aren't visible at all unless I announce them. And even if I announce them, not everyone is listening. And everyone certainly can't listen to everyone else. In some cases, people and firms keep their expectations hidden on purpose (i.e. my employer might tell me about the firms expectations, but if I announced those expectations I expect I'll be unemployed).

Markets are giant muddle. So I don't think it's a matter of people being idiots so much as we're all just finite. We have finite computational abilities to apply to utility maximizing, we have finite access to other peoples doings and plans for doing. Even if we had all the information, we could never process it all. We need to actually do and not spend all our time computing and collecting information about other people's doings.

[rsj]

But you are trying to have it both ways. To play in the big leagues, you need to intimately understand the institutions and be backed by a solid predictive empirical track record.

To engage in poetry, you can play with models and retreat to "the model is not a map". But the problem with Lucas is that he is giving policy advise about lazy workers and too many taxes that is no more useful than that of the Tea Party. And he is hoping that his skills as a poet will give him weight in influencing government and central bank behavior. At best, this shows incredibly arrogance, but at worst, it is an attempt to deceive.

[jamie_2002]

"The map is not the territory" is a truism, but it poiints to some important insights.

Why do we use maps? To help us move around the territory.

Why do maps simplify the complexity of the territory? To focus on the relevant features for moving around the territory.

Does Lucas '72 help us move around the territory? No.

Does Lucas '72 focus on the relevant features for moving around the territory? No.

Lucas '72 is useless as a map.

A useless map does not lead to good directctions. When Lucas and others say, "This is where we should go" based on a useless map they are givng useless directions.

The defence of Lucas '72 in effect is that it is hard to make good maps.

That does not excuse the lack of quality in the map. Nor does it excuse Lucas providing useless directions.

[Phil Koop]

If you feel the urge to leap to Lucas' defense, please consider identifying the assailant. Or at least, the mode of attack. Lucas has suffered some recent criticism for a handful of ill-considered interview remarks, but not particularly for expectations and money neutrality. What amidst these rather banal generalities applies specifically to Lucas 72, and how?

[david]

Nitpick:

"The people in New Keynesian models are complete idiots for waiting for the Calvo fairy to give them permission to cut prices in a recession."

Or, you know, rational reasons for nominal or real wage rigidity might exist.

There is no a priori reason to believe that mass individual rationality necessarily results in aggregate efficiency, or that aggregate efficiency necessarily implies mass individual rationality.

[Nick Rowe]

pe: Someone here posted the link to that paper a couple of weeks back. It looks really interesting. But I just don't get it. I haven't worked through all the math, but I get a really uneasy feeling about their results. Either there's something in the intuition I'm missing; or their results are just wrong. There's an upward-sloping labour supply curve, and downward-sloping MPL curve, as far as I can see, so there ought to be a unique full-information equilibrium. If everybody has pessimistic beliefs, then when 2 islands meet, both should be pleasantly surprised, and learn. I can't figure out that model.

But yes, it does look like they are at least trying to do something like I would want to do.

Kevin: OK. But one of the "rules" is that the model must be simple enough for economists to understand what's going on, and yet too complex for ordinary people to understand what's going on. If economists were 2 orders of magnitude smarter than anyone else, that would be easy.

Patrick: I think you could interpret Lucas 72 in that sort of way. He just assumed that people had very limited information, but were smart in how they used that information, in a simple world, where it wasn't hard to be smart, if you had enough practice by trial and error.

rsj: recessions are not caused by outbreaks of laziness in Lucas '72, and there are no taxes there either.

jamie: I disagree with your assertions.

Phil: the trigger was recent posts by Paul Krugman and Brad DeLong. But the trigger is irrelevant really. I have thought the same thing. All of us have. The assailant is myself, as much as anyone. But now I want to present the other side.

david: agreed. But if you take the NK model literally, there are no such reasons.

[Min]

"All models suffer this same problem. If the world really were as simple as the economic model of that world, people would figure it out, and wouldn't let bad things happen."

1) Not if the model includes suboptimal equilibria.

2) Not if people are not Kantians. Kant would cooperate in the Prisoner's Dilemma. His categorical imperative says to act as if one's actions were universal law. Kantians are not homo economicus. Most people are not Kantians, but they are not homo economicus, either. :)

[Min]

BTW, I ran across this quote yesterday, which seems to fit:

"The sciences do not try to explain, they hardly even try to interpret, they mainly make models. By a model is meant a mathematical construct which, with the addition of certain verbal interpretations, describes observed phenomena. The justification of such a mathematical construct is solely and precisely that it is expected to work."

—John von Neumann

[Stephen Gordon]

I hope none of my students are reading this. I'm teaching the Lucas model tomorrow.

Yes, really.

[W. Peden]

Min,

That is one reason we might model, but it is not the only reason. Very often, we want scientists to tell us about the way that various causal connections work in the world. Galileo's Tower is not a great idealization because it's useful, but because it's an important stage in understanding the causal dynamics of the Copernican cosmos.

[rsj]

No, in Lucas 72 recessions are caused by confusion of real versus relative prices, but Lucas is not being assaulted for Lucas 72, or any of his poetry.

He is being criticized for the right-wing political economics that (he argues) is justified by the poetry. He is being criticized for believing that his poetry has some relevance for policy.

[rsj]

And really, Lucas 72 is a great example of what is wrong here. If you wanted a model of short-tun demand fluctuations -- and his model is only a short run model -- then an engineer or scientist would look at the beast in nature, and would say something like: firms don't want to hire workers because workers can't spend money on firms, or "it is harder to sell than to buy", or something similar.

And then they would try to model something like that.

But Lucas decided to model shifts in the labor supply curve. Whatever insights such an approach can yield, they wont have anything to do with short-run employment and output variations in a recession. There is a lot of data out there to support that view. Lucas didn't look at the data, and he didn't even look at the phenomena that he was purportedly modeling. Instead he looked inward, to his own politics and worldview, and created a model based on that, not based on the world.

He wasn't drawing a map of the landscape, he was imagining his own landscape and then drawing a map of that.

[rsj]

"I hope none of my students are reading this. I'm teaching the Lucas model tomorrow."

Why not read it? The model is valuable because of the analytical techniques that it introduces. Plus, it is beautiful poetry. There is nothing wrong with describing it this way and teaching it, while pointing out that it has no explanatory power in and of itself.

It could be a building block used in some future model that does have explanatory power. For example, if we switch things around and say that investors are confused about the returns that they can expect to get.

Lucas was a great developer of mathematical techniques for economists, and my impression is that he is a great teacher of these techniques. As a non-economist, his papers are among the easiest for me to read. They are clean, clear and very well written. His book is wonderful. He deserves accolades as an educator and mathematician, even if he should be kept far away from any policy levers.

[W. Peden]

rsj,

Not to mention the problems he raised for the more hydraulic approaches to economics. One can make the argument that economics, prior to the work of Lucas, was full of stuff that wasn't a legtimate social science at all but rather just statistical headbanging* with aggregates and adaptive expectations. Even Friedman, who often gets a pass on this kind of stuff (because, I assume, of the permanent income hypothesis and the natural rate of unemployment) still had a view of exogenous money where it was past changes in the quantity of money that mattered, rather than agents' expectations of the future monetary conditions, that mattered. Such a view is perhaps legitimate if one assumes a world with no language, no theories of other minds and in general a world where humans would be too animalistic to have create a monetary economic system in the first place!

My worry about Lucas's models is not that they are "poetry", but that too often their results are assumption-dependent rather than principle-dependent i.e. they depend on assumptions built into the model about agents/insitutions/etc., rather than on the principles that the model uses. However, this is why I agree with you on his writing: mathematization and clear formalisation means that one CAN see that it's the assumptions doing the work, rather than the principles. For this, if nothing else, Lucas is one of the greatest economists of our time.

* The more you do it, the more stupid you get.

[rsj]

I would say that economics before Lucas was more focused on phenomenology. The problem is that even now, say in physics, phenomenology is a valid discipline. Perhaps the successor in economics would be econometrics, but its hard to do theory-free econometrics.

People observed that if you give households more money than they will spend it. They observed that firms, when hiring, don't really think that their workers will bring in revenue simply as a result of being hired. Rather the firms estimate how much they can sell at a given price, and then hire as many workers as are needed to produce that quantity. If you just go to store and ask "how much additional revenue will hiring this accountant bring to you?", then answer would be "zero." And yet the accountant is hired. If you asked, given a certain market demand, how many accountants would you hire at a given price, then the firm will give you an answer. etc.

I think observations about behavior like the above would be a great foundation for economics, rather than attempts to model the human brain -- e.g. quantify "satisfaction" -- and then derive policy advise from that.

Basically, interview people, ask them how they make decisions, ask them what is important to them and how they go about getting it, and then build a model rooted in this observational data.

That necessarily requires that you revamp your model as culture and institutions change, but that's why economics is a social science rather than a physical science. There are no rules that exist in a vaccuum, that is, outside of society. The "Lucas critique" pretends that we have some brain model and can determine how people will respond to any circumstance. It should be clear by now that if some economic principle is not subject to the Lucas critique, then this principle is both useless and untestable. Economics is a social science, and as the society changes, so will the response of individuals to certain policies.

And at the same time, what is so striking in the *policy* arguments presented by Lucas or Cochrane is how childish they are. Right-wing non-economists are able to present much more coherent and compelling arguments. The reason, I believe, is that the former group has intentionally purged themselves of institutional and cultural understanding. This is why their advice seems so abstruse and other-wordly; so far removed from everyday experience, or how firms, workers, and consumers, and savers actually behave.

This can't possible be good when you are trying to "draw a map" of the landscape in which we live.

[Patrick]

To steal a phrase from Wolfgang Pauli: models based on self-reported preferences and cognitive processes? Not even wrong.

Sure, observe what people actually do, figure out the heuristics people use to make decisions, how they cope with limited information, observe what people's preferences really are, find ways to better model relationships between people, between people and institutions, between institutions, etc etc etc...

But just asking them? Not going to work, because they mostly don't have a clue how or why they do what they do.

[Sandwichman]

"It's unfair to single out Lucas '72 for this criticism."

Are you implying that it IS fair to single out someone else -- but not Lucas -- for the same kind of criticism? No, I'm serious. It seems to me that quite interesting models get brushed aside as not worth considering on the grounds that they are, in essence, "insufficiently Lucasian."

"The labour market is even more complex than the housing market."

No shit, Sherlock? So why then did Ronald Coase expend exactly 0 sentences in "The Problem of Social Cost" discussing Part III of Pigou's Economics of Welfare, "The National Dividend and Labour"? And why after Donald Stabile pointed out the anomaly in 1996 has there been no subsequent attention to it from macroeconomic modelers?

That the "labor" in Lucas's production function is manifestly a bad joke should be evident from a reading of Pigou's part III. But of course Coase dispensed with Pigou by dispensing with only part II, didn't he?

[Nick Rowe]

Min: yeh. I slid over a lot of stuff about what "suboptimal" might mean. But if it's suboptimal, but nothing can be done about it, is it really suboptimal? What I had in mind were models where better AD policy could make things better. But if the model were simple enough for an economists to model, we wouldn't need markets, and so wouldn't need AD policy. Central planning would work fine.

Good luck with it Stephen. I wounder what the reaction is of today's students?

rsj: I'm not talking about what Lucas said last week. And Lucas 72 is being attacked, not just in the last couple of days, but in the last couple of decades, by many economists, including me, for assuming agents are idiots in not knowing there's a recession on.

If you put the nominal wage on the vertical axis, not the real wage, or if you put the *perceived* real wage on the vertical axis, then the Lucas 72 recession is caused by a fall in the demand for labour, not a fall in supply.

Yes, Lucas 72 misses the important (to me) fact that it gets harder to sell stuff in a recession. In his model, everything is always perfectly easy to buy and sell at the market price. But Lucas 72 is an outgrowth of the Phelps volume, which included search models. I have always understood it as an allegory of a search model. The relative price could be the price at which you can sell your house or labour right now, relative to the price you could get if you rejected this offer and waited for a better one. Maybe you've just been unlucky, and a better offer will appear if you wait just a couple more days or weeks? Does the slope of that trade off, as well as the intercept, change in a recession? When you and all the other buyers and sellers are making mistakes? "False trading" changes the equilibrium itself. If some sellers initially make a mistake, and don't lower their prices quickly enough, that means a bigger inventory of unsold houses the next period. It might get harder to sell stuff, and easier to buy, in a model like that. Dunno. Too hard (for me) to model.

[Mandos]

"The winners could compensate the losers, under the deal."

Um, ah. Somehow this utterly astonishing remark was made...and then simply passed over by the discussion. Does anyone at all understand why this remark is astonishing? Anyone look askance at it?

[Sandwichman]

"winners compensating losers" is contrary to the motivational structure posited for representative agents. The solution violates its own model. "And then a miracle occurs..."

"In the absence of transaction costs" all kinds of wondrous things occur that can't occur simply because their happening would have depended on the actions of agents who are motivated by the incentive of receiving fees for conducting transactions. That might be the definition of an entrepreneur (let me think about it).

Incentives matter... except when it's convenient to assume the sudden disappearance of those same incentives (or the motivational structures that respond to them) to resolve some hitch of the model.

It's turtles all the way down.

[Sandwichman]

Or call it peek-a-boo rational maximizing. Representative agents are rational maximizers of utility... except when they're not.

[jb]

"The winners could compensate the losers, under the deal."

This sounds like Kaldor-Hicks efficiency.

But in reality winners seldom do..

[Sandwichman]

"But in reality winners seldom do.."

No, jb, that is true but beside the point. The point is that in the model they can't. The proposition contradicts itself. It's a liar's paradox type of proposition: "this sentence is false."

[Frances Woolley]

The Squirrel's Nest's reaction to this post: The Smartest Economist on the Planet.

[Nick Rowe]

Frances: blush! Ah well, there are plenty of other people to provide a balanced perspective.

[Off-topic: do Canadians and Americans, like Brits, call squirrel's nests "dreys"? Is that specialised word disappearing? Through urbanisation, or ESL?]

Not sure if Sandwichman is saying the same thing I would say. If there are no transactions costs in the model, and if the model were literally true, the winners would compensate the losers. "Transactions costs" is a portmanteau term to cover all the things like I was talking about in my housing market example.

[Scott Sumner]

Nick, A few comments:

1. Excellent post.

2. I'm really surprised that the 1982 recession turned you against Lucas. I recall that the 1982 recession was PRECISELY the stylized fact he was trying to explain. At least in terms of prices and output. Maybe I missed something.

3. I was interested in your suggestion that we don't even know what tight money is. Lucas defined it as changes in the money supply. But his model was severely criticized because that data was publicly available on a weekly basis. But what if the stance of monetary policy is changes in NGDP relative to trend? That data comes out with a very long lag in the US. Maybe his model could be resurrected.

4. We basically know people are idiots because of the huge discontinuity in hourly nominal wage increases at zero percent. In economic theory, zero in nominal terms is a meaningless number. In the mind of the public it has a powerful, almost mystical meaning. To me, that's the best argument against Lucas.

[Nick Rowe]

Scott: thanks!

2. You're an economic historian; I'm not. There was really nothing new in the 1982 recession at all, but the experience of actually living through one, with something like the Lucas model at the back of your mind, caused the problem. "Why is this thing taking so damn long to end?? Everyone surely knows by now there's a recession on, and the Bank of Canada wants lower inflation, and so inflation ought to come down immediately!"

In the 1987 stock market crash, I made the point of listening to the news radio all day, and reading old books on the 29 crash, just so I could feel the fear, even though I had no money in the market.

3. Even if "monetary policy" is unobservable, the recession is observable (with a short lag). There's just too much aggregate information out there. Plus, even if "monetary policy" means "expected future NGDP", the average agent knows his own expected future NGDP. (I think that's right.)

4. Yep. I would generalise it. Focal points (like zero) matter.

[Peter]

Nick, I disagree "the recession is observable". There is too much aggregate data out there, but rarely does it come across as information. Fairly recently I recall news articles about confused economists who were pointing at data claiming the recession was over so why did people think it wasn't.

It's a bit off topic, but do the agents in the model ever truly understand the model before the assumptions / calculated equilibrium change? I fully agree that the real world is at least 100x more complex than the model. (What if the person is trying to sell her house because she is considering a job offer in a different city while at the same time the newspaper says they are in a recession!) In many ways this reminds me of an earlier post on never reaching equilibrium because of technical changes (the Mazda Bolt post).

Along the lines of trying to assess whether or not I (we?, my province?, my country?) am in "disequilibrium" I have never understood why the newspapers report the stock market on a daily basis, (or why the default graphs is only 5 days). Is the change in the TSX relative to yesterday a meaningful indicator of anything? Does it have any impact on whether or not I should accept a job offer? Who does this "economic data" benefit?

To address your point about why don't agents move to the proposed solution, if the market was purchasing my product for $10 last year, why wouldn't I try sell it for $10.50 this year? Who cares about "inflation" over-all? I'm interested in what the market will pay me for my product. If it wont sell, then okay I'll lower the price, but not before then. And as long as people keep paying my higher prices, I'll grudgingly accept that the prices on goods I buy keep going up too.

[Sandwichman]

Nick,

No, I'm saying that the "no transaction costs" assumption is a counter-factual whose sole purpose is to call attention to the importance and ubiquity of transaction costs. If wishes were horses in the model and the model were literally true, then beggars would ride -- and winners would compensate losers. But the purpose of the ditty is not exploring alternative transportation options for the lumpen-proletariat. It is a cautionary tale about the futility of day dreaming about pie-in-the-sky. There are no literal situations in which wishes are horses, transactions are costless or winners compensate losers.

Coase was calling attention to the fact that the Pigovian "prima facie case" for government intervention evaporated if you explicitly applied the same frictionless assumption to market transactions as the Pigovian tradition had implicitly applied to government intervention. "In the absence of transaction costs..." and with well-defined property rights the market is the best of all possible worlds.

Was Lucas really telling a cautionary tale about the fatuity of constructing models that abstracted so much from the real world that policy makers "won't have a clue what's really happening and what their optimal response should be"?

[Mandos]

Can anyone tell me of cases where winners compensated losers for any reasonably large classes of winners and losers---without the threat of communism or something? Is this likely to happen any time soon? Where do the losers from e.g. global capital liberalization (trade, finance) sign up?

"Assume a communist..." Communists as can openers! Ha.

[Sandwichman]

"Assume a communist..." Communists as can openers! Ha.

Don't laugh. Hicks assumed essentially the same thing about optimizing the hours of work for output, "as soon as Trade Unionism has to be reckoned with at all seriously…" Seriously. And here we are today with economists (Lucas, for example) implicitly making that assumption without knowing they're making the assumption -- without realizing that they MUST make that assumption if their models are to crank out the results.

[Sandwichman]

as soon as Trade Unionism has to be reckoned with at all seriously…

http://tinyurl.com/jrhicks

[Min]

Nick Rowe: " I slid over a lot of stuff about what "suboptimal" might mean. But if it's suboptimal, but nothing can be done about it, is it really suboptimal?"

Sure. I get your point, though. We may be in limbo, and have to accept our fate. In such a case, why not consider it heaven? "The mind is its own place," as Milton's Satan says. :)

IMO, human systems are chaotic, and so you can't really talk about equilibria. But you can get stable situations that can last for a very long time.

Now, if by smart you mean that people get the gov't to act, well, sometimes they do and sometimes they don't. If you mean that they do something like the spiritual stimulus of Pastor Tim Lucas (I kid you not, that's his name. ;)) ( http://www.onenewsnow.com/Church/Default.aspx?id=1442714 ), well, sometimes they do and sometimes they don't. On a large scale, I don't think that people can act effectively to get out of a suboptimal equilibrium in a short time without organization to form a coalition. Culturally, moral education helps people to avoid or get out of suboptimal equilibria. But I didn't think that economists called that sort of thing smart. I thought they called it irrational.

[Determinant]

I am terribly bothered by economic models that assume the equality of employer and employee. That isn't true, empirically. In an employment relationship an employee has to work to satisfy at least his basic consumption; an employer has to expect that the additional employee will result in sufficient additional revenue to meet the employer's return expectations. Of course employees are also costs and uncertainty in return diminishes expected return. If the employer's return expectations are high enough that they are greater than the risk-adjusted expected return from an additional employee they will often as not NOT hire an employee.

It means that practically an employer has much more power and discretion than an employee does. The ironic thing is that labour unions actually make economists' assumptions about wages being the "marginal product of labour" true, labour unions have the bargaining power to achieve those kinds of wages in contract negotiations. Individual unrepresented employees don't have that power in the presence of the threat that their wage demands will result in their termination. Employers use their power to appropriate employee's productivity for themselves.

I find it funny that unions are implicit in economic models yet most economists don't like them as "monopolies".

[Patrick]

Determinant: fair enough, but I don't think that's what the Lucas model is getting at - though I obviously I don't really understand it in depth. Regardless of how it was received back in 1972, to my completely amateur eye, and to the extent I can decipher in an hour or so of slogging through math I'm not entirely comfortable with, it basically looks very much like one of Nick's thought experiments, fancied-up with lots of math. Just look at the first assumption in the model: production is consumed instantaneously and can't be saved. The imperfect information side of the model looks interesting, but apparently Lucas was really trying to show that money was a 'veil'. Given that assumptions, it's seems a bit much to declare 'Aha! Money is veil' - of course it is. The model is constructed to make it so. But I really don't know anything about this so stuff so I could be talking total gibberish.

[Nick Rowe]

Patrick: it's a model in which: money is a veil; but when the veil flutters, the economy stutters. People have imperfect information, and can't distinguish if the price changes due to a real shock (to which they want to react) or from a monetary shock (to which they don't want to react). So they react "half"way, only it's not half, but depends on the real signal to monetary noise ratio. Optimal monetary policy is to minimise monetary noise. If it's doing that, it doesn't matter what monetary policy does.

You are very brave to work through the math.

[Nick Rowe]

Peter: suppose you are already choosing your own optimal best price for your product, and everyone else is too. So there's an equilibrium, and we're in it. Then if the money supply expands 10%, there has to be another equilibrium where all our optimal prices rise by the same 10%. And if we all know that, and know everybody else knows that, etc. (if it's common knowledge), then we will all immediately raise our prices by 10% and jump immediately to the new equilibrium. But what happens when it isn't common knowledge, and we are all learning at the same time, is the tricky question. Lucas gave one answer to that question. It was a complicated answer, even though it was a massively simplified model.

[rsj]

Hmm, let me translate this into simpler english:

Everyone is 6 feet tall and they demand to live in house that is 8 feet tall, so that is the equilibrium: 6 foot people in 8 foot houses. Now suppose the height of houses spontaneously increases by 10%. Obviously the height of people will need to rise, too.

Unless people are confused about the height of their houses. Perhaps it was their own house that spontaneously got bigger (without them doing anything). Or maybe all the houses are getting bigger. Who knows?

These were the types of of nonsense questions that Lucas was struggling with. "Spontaneous increase in the money supply"? How can something that is demand determined increase spontaneously? Ex-nihilo, you look in your wallet and there were $100 when before it was only $50? And then how would people respond to this fantastic event? What insights are we to gain from such disputations?

[Patrick]

Thanks Nick.

[Sandwichman]

"Everyone is 6 feet tall and they demand to live in house that is 8 feet tall..."

This can't be true!
I'm six foot two.

[Min]

rsj: "Everyone is 6 feet tall and they demand to live in house that is 8 feet tall, so that is the equilibrium: 6 foot people in 8 foot houses. Now suppose the height of houses spontaneously increases by 10%. Obviously the height of people will need to rise, too."

;) ;) ;)

[Peter]

Thank you Nick, you're going to be sending me back to the textbooks for sure! (And I thought I was done with macro... Too bad I don't get to personally increase my money supply 10% to see what it does to my equilibrium.)

[Robert Waldmann]

Warning. This comment is rude. I have been as polite as I can force myself to be.

I reject your alleged proof. You assume that a benevolent social planner is available.

I think that market outcomes are inefficient. I don't want to appoint a dictator. I may or may not be an idiot, but the conclusion that I must be an idiot doesn't follow.

Consider crime. We don't like crime. We ban it. We have police and stuff. We have crime. This isn't because we are idiots. There are two reasons. there are enforcement problems with command and control solutions (even if the command is "don't kill each other"). Also dictators are dangerous. We can't count on their benevolence.

Your assertion doesn't even follow in the imaginary world of Lucas 72. Lucas's model does not include a benevolent agent who is able to make suppliers supply. It isn't there. You have modified the model fundamentally changing it, then assert that the result for your new model holds for Lucas's model.

I have two questions

Do you seriously believe what you typed at the beginning of this post ?
and
How could anyone type such a thing ?

Your conclusion, that models are misleading just because they are (and should be) simple and the world is complicated is a very good point. This is definitely true. It is important. The always implausible hypothesis of rational expectations is much more appealing if the model with which expectations must be consistent is simple enough to solve.

I think this would be an excellent post if the first seven sentences were deleted (so it would start with "The people in Lucas'"). But those sentences are nonsense.

As an aside, nominal rigidities and price prediction errors are not at all needed for Pareto inefficient outcomes. Market outcomes can be Pareto efficient even if people have perfect foresight and all markets clear (consider pollution externalities).

To belabor the totally blindingly obvious fact that your assertion is false.

You claim that, any day we wish, we could have, for example, perfectly efficient regulation of polution. Just walk down any street (in the USA and Italy anyway) and see how there is no litter. All at the low low perfectly efficient zero cost of enforcing anti littering laws. You confidently assert not only that there ought to be a law, but also that the law will work perfectly (not well perfectly) unless the legislators are idiots.

I ask for information. How could you have written something so obviously false ?

[Stephen Gordon]

I'm pretty sure that people who are using this post as a springboard for righteous wrath are missing its point.

More generally, anyone wishing to use WCI as a springboard for righteous wrath are invited to jump in other parts of the internet. It's a big place; there are plenty of outlets for it.

[Stephen Gordon]

*sigh*

Last comment was not an invitation to explain why your righteous wrath is justified.

[Nick Rowe]

Robert Waldmann: "I have two questions

Do you seriously believe what you typed at the beginning of this post ?
and
How could anyone type such a thing ?"

I have the image of an irate schoolteacher glaring down at a 5 year old.

Knock it off. I can't stop you acting like that on your own blog, but as Stephen said, it's totally out of place here.

And you might want to reconsider the whole irate ursine act in general. History is littered with angry people doing stupid and evil things because they believed anyone who disagreed with them was stupid or evil.

And thinking about that, and your comment, makes me angry too.

Perhaps what I said at the beginning of this post isn't as clear as it should be. Perhaps it's wrong. But stupid?

Suppose you have a model in which bad things happen, and you think things could be better. I can think of three cases to which all such models might ultimately be reduced:

1. Prisoners' Dilemma. So, what are you going to say? "If you guys would only cooperate, you would both be better off"? If they could cooperate, it wouldn't be PD.

2. Stag Hunt. Again, what are you going to say? "When I give the word, both of you switch from hunting hare to hunting stag." Couldn't they say that themselves?

3. A maze. You, as modeller, have some sort of bird's eye view, and can see the route out of the maze, which the players themselves can't. OK. Maybe in this case you can say something useful. But if the maze is simple enough for you to model, it ought be be simple enough for the players themselves to model, and figure their own way out.

My gut says that the maze might be the best way to think of what economists do, as an advice-giving profession. After all, someone's got to be the first person to figure out the maze. And it might be one of us. But there's still a tension here. If the maze is literally as simple as our model says it is, how come the people stuck in the maze haven't figured it out already?

[Steve Roth]

Yeah, of course. But some modeling approaches are better at their jobs than others.

"Start in equilibrium (whatever that means in a housing market like I've just described)."

Imagine if climate scientists used static "equilibrium" models like this. Sheesh. Is a warm sunny day at some point on the earth in "equilibrium"? A rainy day? A blizzard? How about 100 miles away? Kind of a meaningless question. Arguably a useless approach to modeling. (Hell, we don't even know what direction the IS curve slopes.)

The only working dynamic model (assuming there is no imagined "equilibrium," only constant change) that I know of is Steve Keen's. You can download it and run it. (He's developing a new version, web-based.) It's pretty simple compared to climate models, but it seems to have some predictive power, and it certainly has some illuminative power.

http://www.debtdeflation.com/blogs/qed/

Some guys at Santa Fe Institute, last I heard, were "hoping to" develop such a model. I hope they put it on the web where we can run it through its paces.

Shouldn't there be dozens, hundreds of top-flight people trying to build these models? I corresponded with a young PhD candidate who wanted to work on one, he said *all* the institutional systems conspired to prevent it. "Just grab a data set and throw some regressions at it. Build a static equilibrium-based model to theorize the results. Collect PhD."

[Nick Rowe]

Steve Roth: you have been *very* badly misinformed. (By who?) Google "DSGE".

[Robert]

Calling a model "dynamic" doesn't make it so. Here's some overviews of DSGE models:
http://cscs.umich.edu/~crshalizi/notabene/dsges.html
http://robertvienneau.blogspot.com/2010/11/slight-illness-doctor-jests-king-today.html

Nick Rowe would have done better to refer to Agent-Based Models (ABMs). I like some work Ian Wright has done with them.

[Nick Rowe]

Robert: they are "dynamic" in the sense of having an equilibrium in which the endogenous variables move over time, as opposed to a "static" equilibrium. And, DSGE models are common. ABMs are relatively rare. Since Steve was talking about what economists do, DSGEs give a better typical picture. But yes, one could add ABMs as well.

I expect you could make a case that ABMs are one way to try to tackle the main point of this post?

[Robert]

The Arrow-Debreu model of intertemporal equilibrium is a static model in which all agents have their plans for the future pre-coordinated at a single instant of time. Despite the model being static, relative prices differ for forward trades of different dates.

Robert Shiller pointed out in 1978 that RBC modelers don't take seriously out-of-equilibrium behavior and the question of how equilibrium is established. As far as I know, Shiller's point remains true for DSGE modelers.

Steve Roth is correct in thinking that mainstream economics typically suppress modeling of economic dynamics.

What point? I agree with Robert Waldmann that it is silly to think that just because a model shows a suboptimal equilibrium, the agents must be either idiots or be able to establish an optimal equilibrium, perhaps with a dictatorship.

[Nick Rowe]

Robert: "As far as I know, Shiller's point remains true for DSGE modelers."

I think that is correct too. Trouble is, what does "equilibrium" mean? It doesn't mean "point of rest"; it doesn't mean (or have to mean) "supply=demand". If you had a model which showed how (say) P and Q moved over time from one supply=demand equilibrium to another, wouldn't you want to say that the time paths of P and Q in your new model were an "equilibrium" of that model? In other words, "equilibrium" means "whatever the model says will happen".

In an ABM, "equilibrium" means "the time path traced out by your computer"?

We might follow Hayek, and define "equilibrium" as when agents plans are mutually consistent. And "disequilibrium" means agents are changing their plans because they discover their plans are mutually inconsistent. But even here, if agents make a contingent plan, depending on what others do, they aren't really changing their plans, just following different paths of that plan, according to how events unfold.

[Nick Rowe]

Robert: "What point? I agree with Robert Waldmann that it is silly to think that just because a model shows a suboptimal equilibrium, the agents must be either idiots or be able to establish an optimal equilibrium, perhaps with a dictatorship."

Then you are missing my point. At the risk of having more enraged people piling on again: my point is not about the world; it is about modelling the world. Yes, bad things happen. And we should try to make them better. Maybe with the help of a model. We have models showing bad things happen. Does the model show how to make things better? If it does, then why haven't the agents already figured it out themselves? Maybe it's too complex for them? But if your model is simple enough for you to understand, and if the model were literally true, why can't they understand it, unless they are less smart than you? Maybe the answer is that models aren't literally true and are simpler than the world they describe, so it's easier to figure out what's wrong in the model than to figure out what's wrong in the real world?

[Robert]

"If you had a model which showed how (say) P and Q moved over time from one supply=demand equilibrium to another, wouldn't you want to say that the time paths of P and Q in your new model were an 'equilibrium' of that model?"

No, not necessarily.

"We have models showing bad things happen. Does the model show how to make things better?"

Often, no.

[nathan tankus]

@nick rowe

i think what you are missing is that agents don't necessarily have the ability to optimize the functioning of a system, or at least an economic model doesn't need to make that assumption. only from making this assumption can you come to the conclusion you have come to.

let's take pareto optimality. there are two major problems with this, first losers don't necessarily have the ability to enforce compensation from winners. second, only in a model that reaches equilibrium where agents have perfect information is this possible. let's take my importer example from earlier. even if he has a mechanism to enforce compensation payments, how much can he demand and how much would actually compensate him. in a non-linear dynamic world, how does he know how much losses import quotas will put on him? would the winners benefit overtime as much as he loses? how do we know that? how is it even possible to calculate?

[D. I. Harris]

Allow me to play Devil's Advocate for a moment: is the "people are stupid" explanation really so wrong? After all, economists also make stupid mistakes: the ECB is bragging—bragging!—about how the Euro only experienced 1% inflation while in the midst of a debt crisis. The solution is simple and obvious: if the ECB targeted 5% inflation every year for the next 5 years, the debt crisis would be mostly over (Greece would probably still be in trouble, but that's it).

But it is not done. The debt crisis could be over, if the ECB wanted it to be, but it doesn't. Considering the bragging it did, stupidity seems the only option.

So is it any surprise that regular people have difficulty doing the easy solution, when the ECB can't do it either?