The headline summarizes the observations of economist Paul De Grauwe, who takes central banks to task for their reliance on so-called Dynamic Stochastic General Equilibrium models (DSGE). De Grauwe objects to some of the fundamental assumptions embedded in them (consumers are rational and all have the same preferences, any disruptions are the result of external shocks, as opposed, say, to internal imbalances, such as misallocation or mispricing of capital).
Another issue that appears to be implicit in these models (and the consequences have been discussed in more detail by economist such as Axel Leijonhufvud, Richard Alford, and Tim Duy) is that the Fed views the economy as a closed system and has not made sufficient allowance for the impact of trade, particularly how cheap imports have led the central bank to misread domestic inflation (ie, excluding the trade sector) and adopt overly lax monetary policies. It seems that this mis-framing of the problem might have been aided and abetted by reliance on DGSE models.
Now it’s a given that any model of a system as complex as an economy is bound to have some shortcomings. But when analyses have biases and limitations, the best approach is to use multiple methodologies and use judgment and empirical cross-checks. Over-reliance on a particular methodology too often leads users to unwittingly default to it.
From De Grauwe at EuroIntelligence:
One of the surprising developments in macroeconomics is the systematic incorporation of the paradigm of the utility maximizing forward looking and fully informed agent into macroeconomic models. This development started with the rational expectations revolution of the 1970s, which taught us that macroeconomic models can only be accepted if agents’ expectations are consistent with the underlying model structure. The real business cycle theory (RBC) introduced the idea that macroeconomic models should be “micro-founded”, i.e. should be based on dynamic utility maximization of individuals. While RBC models had no place for price rigidities and other inertia, the New Keynesian School systematically introduced rigidities of all kinds into similar micro-founded models. These developments occurred in the ivory towers of academia for several decades until in recent years these models were implemented empirically in such a way that they have now become tools of analysis in the boardrooms of central banks. The most successful implementation of these developments are to be found in the Dynamic Stochastic General Equilibrium models (DSGE-models) that are increasingly used in central banks for policy analysis. It is no exaggeration to say that today a central bank that wishes to be respected has to have its own DSGE-model.
These developments are surprising for several reasons. First, while macroeconomic theory enthusiastically embraced the view that agents fully understand the structure of the world in which they operate, other sciences like psychology and neurology increasingly uncovered the cognitive limitations of individuals. We learn from these sciences that agents only understand small bits and pieces of the world in which they live, and instead of maximizing continuously taking all available information into account, agents use simple rules (heuristics) in guiding their behaviour and their forecasts about the future. They do this not because they are irrational, but rather because the complexity of the world is overwhelming. In a way it can be said that using heuristics is a rational response of agents who are aware of their limited capacity to understand the world.
A second source of surprise of the development of macroeconomic modeling in general and the DSGE-models in particular is that other branches of economics, like game theory and experimental economics have increasingly recognized the need to incorporate the limitations agents face in understanding the world. This has led to models that depart from the rational expectations paradigm. DSGE-models have been immune from this trend.
The rational expectations assumption embedded in DSGE-models has another far-reaching implication. When all agents understand how the world functions in all its complexities, there is only one “Truth”. Everybody understands the same Truth. The implication of this extra-ordinary assumption is that one can restrict the analysis to a “representative agent”. As a result, DSGE-models routinely restrict the analysis to a representative agent to fully describe how all agents in the model process information. There is no heterogeneity in the use and the processing of information in these models.
So, what do the DSGE-models teach us about the sources of macroeconomic fluctuations? They tell us a story in which rationality of superbly informed and identical agents reigns. Shocks from the outside occur continuously forcing these agents to re-optimize all the time, which they are eager to do. Unfortunately and inexplicably, the outside world imposes restrictions on this optimizing behaviour. These super-rational creatures would like to adjust their prices and wages instantaneously so as to maximize their utilities but they are prevented from doing so. They have to wait in line to adjust prices (“Calvo-pricing) and as a result they cannot achieve the optimum immediately. Thus, these outside shocks create distortions and departures from optimality. They also create slow adjustment in output and prices. A theory of the business cycles is born. Business cycles arise because exogenous shocks occur. These shocks are transmitted slowly into the economic system, producing protracted movements in output and prices.
Let’s apply this idea to real-life business cycles and let’s concentrate on the US business cycle since the 1990s. Here is the DSGE-story. Around the middle of the 1990s a productivity shock occurs. This raises permanent income of US consumers leading to a consumption boom. It also leads to more inflation and an interest rate hike by the Fed around 2001. After a brief slowdown, US consumption roars further on its upward path. Then suddenly in August 2007 an exogenous and unforeseen shock occurs, i.e. an increase in the financial risk premium. This changes the outlook for rational consumers and the US economy experiences a severe slowdown.
The problem with this business cycle theory is that it is not a theory of the business cycle. In DSGE-models cyclical movements are always triggered by outside shocks. They are not generated within the system. In fact they cannot be. The super-rational and fully informed creatures that populate the DSGE-models would arbitrage away any systematic cyclical movement in prices and output. Thus, our most popular macro economic model that is now used by central bankers has no theory of why after a boom comes a bust. Booms and busts always find their origin outside the macroeconomic landscape.
This is a very unsatisfactory view. The downturn in US economic activity since last year is the result of excesses in the boom experienced before. In this sense the bust is caused by the boom. It is not the result of some new and unexpected financial shock as the DSGE-models tell us.
Macroeconomics is about systematic fluctuations in output, employment and prices. Macroeconomics is also about social interactions between agents who do not understand very well how the world functions. As a result, they watch each other to get clues of what is going on in the world. This leads to herding and group behaviour. This social behaviour is at the core of macroeconomic fluctuations. Keynes gave this a name, “animal spirits”.
All this is absent in DSGE-models where agents who understand the complexity of the world, behave in an atomistic way. There is no need to learn from others, since each individual’s brain contains the full information. Everybody understands the “Truth”.
In his famous AER-article Hayek (1945) stressed that individuals have only very small parts of the available information in their brains. No individual can ever hope to understand and to process the full complexity of the world in which he lives. That’s why markets are so important. They are institutions that allow to efficiently aggregating the divers bits of information stored in individual brains. The socialist economists at the time in contrast assumed that there was one individual, “the planner”, who understood the whole picture, the Truth. By giving him all the power this all-knowing individual could compute all the relevant prices and so force the optimum on the system. Markets were not necessary in this view.
Paradoxically, the rational expectations assumption that is now embedded in macroeconomics created a model that, like in the socialist models of the past, assumes an all-knowing individual, who can compute the optimal plans and set the optimal prices. In such a world, markets are indeed not necessary. We can trust this one individual to do it right.
Macroeconomic modeling should go back to the old Hayekian idea that information is spread among many individuals and that markets are there to aggregate this information, without any individual able to comprehend the whole. This aggregation process is at the core of the macroeconomic fluctuations. Individuals are struggling to understand the complexity of the world. They form opposing beliefs about the “Truth”, but they are willing to learn from others. As a result, we obtain fluctuations in beliefs that have a self-fulfilling aspect and drive the business cycle. The challenge for macroeconomics is to model this aggregation process, instead of assuming it away.
I agree very much with De Grauwe’s criticism’s and most of his observations. Most markets and all economic systems are profoundly open systems; indeed, they are the _most_ open of social systems, as price and availability link participants and sub-systems which have no shared language, religion, government, education level, class background, or any other parameter. The closed system expectation is sheerest fantasy.
To the extent that markets and economic systems are stable, this is more because participants _do not_ all act the same way: to do so leads to positive feedback loops which overshoot and kill the money. And while current macro theory likes to talk about ‘external shocks,’ systemic stability is substantially maintained by external _buffers_; what else, do tell, are the TAF and TSLF? The fantasy that market systems optimize to stability does not survive any engagement with historical facts on the matter. But then the model makers have never cottoned to historical political economists, have they? Pseudo-statistics look good and pay better as long as you’re not playing with your own money.
Even worse is the ‘uniform rational actor’ hypothesis, a verse of Genesis somehow omitted from the Old Testament, and neither more nor less substantive than what all else made it into that primary text. I invite anyone who wants a larger, firmer, and more current perspective to read current work in psychology and neuroscience on categorical perception, to say nothing of studies on behavioral economics. De G. has it a-right, that most market and financial system participants operate by heuristics, which furthermore I would add are categorically formed but which _do not formulate information ‘rationally.’_ That is not to call them ir-rational, but the organization within them is not defined by any logical and naive physical scientistic calculus. Received schematics of how economic systems work needs badly to be updated for what has been learned over the last sixty years on how the mind and the human animal work.
I will say here flatly that the business cycle is endogenous to social systems, but at base it has nothing to do with economic activity. Since I haven’t published, I can’t prove that, so anyone inclined to take my loudmouth at full value on that is welcome do do so; I do however, have good evidence on that issue, so the day for a real discussion may come. Now, I do not disagree with De G. that herd behavior does happen within economic trajectories; in fact, I agree strongly, and such inefficient aggregation is a real and major fact in ongoing economic fluctuations. The background oscillation, though, is non-economic. The whole concept of ‘macro-economic exogenous drivers’ desperately needs to find is place in the trashbin of history; 120+ years and counting it hasn’t worked, nor will it ever.
As a final point, the concept that there exists a ‘God’s eye viewpoint’ from which a perfectly informed actor could read markets or the system is deeply embedded within the logic of naive classical physics which informs macroeconomic theory as we now to often have it. This is because that naive [in the technical sense] understanding, what I call ‘the physical rhetoric’ which is also found in historical reasoning and in social science generally, is at its base simply received theological cosmology with the God-image in the picture reduced to invisible rays; a redefinition of pre-physical natural law rather than a replacement of the ‘logic’ of such thinking. Concepts from self-organizing and complex systems have yet to enter into economics in a significant way, and their relational ‘logic’ is, unfortunately, not yet to be found, certainly not amongst academically trained economists. Those who learned their economics working with money and capital flows may have a heuristic feel for such ‘systemic logic,’ but not a comprehensive perspective. We are still talking about ‘God(s)’s will’ when we should be talking about dynamical mutual modulation because that is the frame of the embedded logical structure in the rhetoric or contemporary economics; consider that the next time someone tells you whether interest rates should go up or down.
. . . We can do better, is all I’m sayin’. But we have to think differently to do so.
“…So, what do the DSGE-models teach us about the sources of macroeconomic fluctuations? They tell us a story in which rationality of superbly informed and identical agents reigns. Shocks from the outside occur continuously forcing these agents to re-optimize all the time, which they are eager to do….”
I see, so the people who bought homes in California in 2005 all paid on average the most appropriate price, the lenders provided the most appropriate level of financing, the investors who ultimately funded the loans assumed the most appropriate level of risk. And further, MR Central Banker, thanks for explaining to me that today’s problems weren’t the fault of any of the market paricipants but rather were caused by an “unforseen and exogenous event”. I feel much better knowing this was really nobody’s fault.
RK said: “…the concept that there exists a ‘God’s eye viewpoint’ from which a perfectly informed actor could read markets or the system…”
You now telling me this ain’t so? I thought this was the central tenant of quant philosophy. I guess, next you’ll be saying there’s no Santa Claus.
“The most serious challenge that the existence of money poses to the theorist is this: the best developed model of the economy cannot account for it.”
— Frank Hahn
Here he is speaking of Arrow Debreuvian models of general equilibrium.
Until economic models begin to treat money, time, and uncertainty (along with financing and asset markets) seriously, as an integral part of all economic activity rather than as a bolt-on, they will continue to have laughably inadequate explanatory and predictive power.
I can’t believe Yves actually writes:
“Now it’s a given that any model of a system as complex as an economy is bound to have some shortcomings. But when analyses have biases and limitations, the best approach is to use multiple methodologies and use judgment and empirical cross-checks. Over-reliance on a particular methodology too often leads users to unwittingly default to it.”
While also having written 114 posts on global warming and it’s model to wit:
Man makes C02. Man creates global warming. End of story. End of model.
I guess climate sysytems are quite simple after all…
As to Mr. Kline’s statements:
“Received schematics of how economic systems work needs badly to be updated for what has been learned over the last sixty years on how the mind and the human animal work.”
“We can do better is all I’m saying….”
Maybe we can do better….But once we do, the schematic that allowed for the improved performace will also need to be updated…on and on the updates will continue… each needing to happen at ever decreasing intervals…until the time between needed improvements is so brief that it becomes instantaneously priced in. A state of constant flux, disguised as equilibrium. A market place of ideas.
Each new schematic will be worth something until it’s commoditized. Then, all we’re left with is an intangible commodity…and we all know how much those are worth. [Wall Street in action…take an idea (“schematic”), reap the hell out of it until its stripped of all value, then package and market it in such a way that the Main Street Buyer thinks he or she is actually a part of Wall Street…sell it to them and reap some more(at this point, it may be more accurate to move the “e” in reap to a place after the “p”), while employing some new schematic that nullifies the old.]
The economists who employed the use of “rational actors” were not being naïve. On the contrary….When a novel concept like Supply and Demand was first expressed “scientifically” as a model, …it took a rational eye to observe whether the model was legit. Then, a clever economist, asked what would happen if we made a model which includes the rational observer.
Now come the New Economists who state not only that the old model is out-dated (which it probably is), but that the old model makers were somehow naïve and simple-minded for creating such a model. “How silly were those economists who actually think man is rational!”
Any New Economist thinker (like Taleb) who flicks away models from the past with disdain, fails to see that the new paradigm (some neuro-science, behavioral finance concoction with “cognition” thrown in there somewhere) does not replace the past paradigms any more than the neo-cortex does away with our reptilian stems. The amusing part is these are the same people who tout evolutionary models, while being oblivious to the fact they are part of an evolutionary model.
There’s a reason for the “God’s eye view point” as you call it, and it’s not because the people or societies who engage in it have failed to read Tversky and Kahneman. As Godel pointed out, in order to prove (i.e. evaluate) a logical system, you MUST step outside of it. No exceptions. Thus, a God’s eye view point is necessary to evaluate any logical system. After it becomes apparent that there will be many “evaluators” it makes sense to incorporate these evaluators in a new model. The old gods are cast down amongst the subjects, while the New Gods evaluate anew. Don’t look to Genesis, rather give Exodus a look.
On the statement:
“I invite anyone who wants a larger, firmer, and more current perspective to read current work in psychology and neuroscience on categorical perception, to say nothing of studies on behavioral economics”
I invite you to read about the sample sizes and methodolgies used in current social sciences….
You’ll read about sample sizes of 5 college dudes who couldn’t get a ride home for Spring Break, so they decided to go to the social sciences lab and make few extra bucks.
Please excuse this most unscientific, generalized statement:
The social “sciences” conduct some of the most biased, shoddy work in academia. It is not science…not even close.
I’m with Dan Duncan. Any mathematical model of human behavior is stupid, but not only because humans can be irrational and uninformed. Humans understand their interdependence and know that extreme concentration of wealth or power is neither “fair” nor “efficient” in any meaningful sense of those terms. The general public has never accepted the notion that important political decisions like the minimum wage, income tax, social welfare programs, etc. can be based solely on mathemtical models. They’re smarter than that. They also have never quite found a practical use for that miracle device economists invented, the only one in history whose market behaves in a predictable way and about which there is perfect information and no transaction costs: the widget.
Complexity, compounded by the ability of some players to alter the rules, coupled with a large dose of ‘opacity’ results in confusion.
Naturally, the problem lies within the ‘rules’ that govern the system.
I’d posit our largest problem lies within the basic assumptions regarding the ‘properties’ of money…that ‘odd’ little invention intended to ‘simplify’ barter that has subsequently failed miserably.
Here’s a little meme for you to chew on. Money serves but one useful purpose to a society, it is a ‘regulator’.
It is the ‘misapplication’ of this one useful function that causes economic ‘shocks’.
Like the incredibly bizarre idea that money can make money. It doesn’t and never will.
Money ‘represents’ wealth and the only way to create wealth is to labor for it.
Here’s my heuristic: market values are set in equal part by fundamentals (supply, demand, growth, productivity, etc.) and by sentiment (human herd behavior which causes systematic, emotionally-driven overshoots and undershoots in valuing goods and assets).
Models assuming a “representative rational actor” are wrong and useless. Economics is not a science, because its fundamental premises are wrong. This may be why not one in a hundred economists has ever correctly forecasted a recession.
Re: “Money ‘represents’ wealth and the only way to create wealth is to labor for it.”
But different types of labor can produce vastly differing amounts of wealth. Compare and contrast: Digging a ditch; digging an ditch and installing a producing oil well there; being in an army; leading an army; digging a cave to live in; finding gold in that cave.
Etcetera, etcetera, and so forth.
neoclassic economic’s bias towards the market slights how and why what is placed into the real market comes into existance, i.e. the process of production and contradictions inherent to it as well as between production and consumption.
IOW, as a process of circulation, the market cannot create one iota of new value though is absolutely necessary for the realization of value. The cycle is a rate of profit and rate of accumulation cycle; it is capital’s incessant drive to overcome itself.
this same economics often fails to avoid ahistoricism as it inclines towards reading the present set of dominant social relations as infinite rather than historically limited. Evidently, discontinuities are not permitted or must be assigned to ‘exogenous’ forces — which is really no more than a means to excuse system and theories. Some might call modern, mainstream economics what it has been, an ideology supportive of a particular class.
From a more mainstream perspective than my little note, the attached dovetails nicely with Yves’ post:
What Is Neoclassical Economics?
The three axioms responsible for its theoretical oeuvre, practical irrelevance and, thus, discursive power
(Arnsperger & Varoufakis, 2006)
The analogy you make to global climate modeling is pretty weak. Yes climate models and economic models both are inadequate representations of the things they are modeling. But the critical difference is this: economic models are invariably proven wrong in short order and err in all sorts of directions. Global climate models are very consistently UNDER-stating the problem. Actual results have been consistently WORSE than the models have predicted.
You can’t write off models as a kind of mass delusion when the real world is behaving in ways that are more extreme than the “delusion” predicts.
To Anon of 9:54 AM: I profess no expertise at all regarding what the financial quants are up to, but that said I doubt that they give a rat’s rictus regarding any God’s-eye viewpoint, or take the idea of ‘perfect information’ seriously. The early systems dynamics types did, but I think the complexity of such analyses is more widely appreciated, now, vide the subject of Yves’ post. What I _suspect_ that the quants are up to involves defining limit parameters for financial variable movements, identfying critical threshold values, and then approximating aggretations around or trend deflections toward those values so they can bet on them. Failing that, they could try for value space distribution analyses to at least find clusters to bet on as they occur. In either case, ‘perfect information’ is meaningless since what is sought is ‘current information’ that is meaningful.
To Dan Duncan: An extra-systemic viewpoint isn’t necessarily a God’s-eye one; it can be simply one imperfect in a different way. That can still be useful, since multiple imperfect views may still supply an aggregate parallax. Consider Rashomon: we still don’t know who of the three was ‘right,’ but we do know much more about who the three _are_, what the range of predicted behavior was, and so have some way of constructing an ‘event space.’ But those who constructed historical reasoning and economic optimality models in the 19th century really were committed to the idea of perfect or near-perfect representation; the whole positivist thing. I could go on, here, but I’ll stop now.
Regarding naive economists of three and four generations ago, I hesitated to use that word because it leads to misunderstanding. I don’t mean naive in the sense of silly or hopelessly uneducated, but more in the technical sense for which we now use the word ‘heuristic;’ unweighted, all things being equal, common sense. A naive view is often in the middle of the bell curve, correct for a limited outcome set but not something you can extrapolate. I think this is much the case of the ‘naive rational actor’ and their putative perspective built into current economic analysis; it is not so much ‘wrong’ as only right for a subset of outcomes, and not something one can extrapolate. . . . Of course economists go on and extrapolate that rational perspective, and that isn’t naive it’s foolish, which is more my point. In fairness to the thinkers who framed economics as we know it in the later 19th century, they did so before Freud’s work, flawed as it was, generated the first real theory of mind as distinct from philosophies of something more like a soul. Those classical economists wouldn’t and couldn’t see behavior in the way we might now. —But we needn’t continue to confine ourselves to their perspective. [Note best not to tag me as RK since someone else commenting here goes by that monicker; try Richard K., or Kline if you must.]
To DD of 11:03 AM: Few could be more critical of social ‘science’ than I am. Manifestly, many ‘studies’ there are poorly constructed. That said, not a few are well-constructed, with well-documented replications, some of whose conclusions tie in with reasonably firm evidence in other fields. It would be easier to throw it all out, but there are gold nodules in the dross. Again, I recommend reading current work on categorical perception, memory formation, and selectively in neuroscience where studies correspond.
To Juan: I completely agree with the points of your last paragraph. The ahistorical bias in the models of contemporary economics, technically a synchronic perspective, is deeply flawed in every regard. One compresses past differences, and expresses present states into the future with insufficient qualification, getting distortion as a result. Better economic modeling is explicitly diachronic in my view, i.e. ‘comparative’ of historical states, even over the short-term of months rather than decades. That is exactly what mainstream economics goes to great lengths to avoid: we see the result. Not surprisingly, actual financial professionals lean as heavily as they can on state comparison, although to their misfortune they’ve been trained to filter that info through classical economic distortions if they went to business school.
I’m with you as well, Juan, that mainstream economics is fundamentally an ideology with a pronounced political overtone. Now, these economists _do_ attempt to do substantive analysis nonetheless, but they have to frame their models through ideology and interpret their results through ideology. No viewpoint is free from biases, but few viewpoints are as bias-dependent as modern economics, neoclassical certainly, perhaps neoliberal a bit less so but it’s a matter of degree not kind. I don’t have a solution other than that the field needs a major make-over.
“Global climate models are very consistently UNDER-stating the problem. Actual results have been consistently WORSE than the models have predicted.”
You’re kidding, right?
Are these the same models created in the 1970s out of concern for Global Cooling? From the same scientists featured in the cover story of Newsweek in ’75 entitled “The Cooling World”? Or how about this nugget from one of the models: “The evidence on support of these predictions on Global Cooling has now begun to accumulate so massively that meteorologists are hard pressed to keep up with it”?
Here’s a suggestion from the some respected climatologists of the day…and no, I’m not making this up: “Possible solutions to the Cooling Crisis posed by scientists are covering the polar ice caps with black soot in the hopes of MELTING these caps”?
Perhaps the accurate climate models to which you refer relate to Newsweek’s correction article some 30 years later: “Climate models have been so spectacularly wrong about the near-term future…”
Of course this denial was followed up by the editorial from Newsweeks’ Editor, Jerry Adler…and again, I kid you not…that the story “wasn’t wrong in the journalistic sense of ‘inaccurate'”
What a joke.
Then STS writes:
“You can’t write off models as a kind of mass delusion when the real world is behaving in ways that are more extreme than the “delusion” predicts.”
Actually, STS, I can.
The point is this:
Either Yves stands by the statement: “Now it’s a given that any model of a system as complex as an economy is bound to have some shortcomings. But when analyses have biases and limitations, the best approach is to use multiple methodologies and use judgment and empirical cross-checks. Over-reliance on a particular methodology too often leads users to unwittingly default to it.”
Or Yves doesn’t.
Either way, the issue deals with Modeling, not economics or Global Warming.
When we apply Yves statement to an issue which might be modeled…like a climate… then, according to Yves methodolgy, we must ask:
Is climate modeling complex?
Does the analysis of climate modeling have biases and limitations (on either side of the issue)?
Is there an over-reliance on a particular methodlogy?
We know these questions are answered in the affirmative. So, I am curious as to how Yves prescription of “judgement and empirical cross-checks” is being applied? Because I don’t think it is. Not by Yves, anyway, because not once does Yves ever present another viewpoint other than Man-Made carbon causes Global Warming…unless, you include the fear that man-made nitrogen is also a culprit.
[Frankly, catchy, feel-good, but statements like “judgement and empirical cross-checks” are too subjective and woefully vague in the face of complex systems with inherent biases.]
The article I cited makes it clear than a single model, the DGSE, plays a dominant role in central bank thinking. By contrast, while the IPCC report did use models (I am not certain whether it can accurately be called a single model, since it appears to have had large sub sections that could be called models in and of themselves) the IPCC does not have monopoly on climate science. Other models by other scientists have come up with broadly similar findings. And as STS indicates, real world observations, particularly Arctic ice cap melting, are happening far faster than the models predicted.
I suggest you read this article from New Scientist, “Climate change: A guide for the perplexed.”
Another way that the climate models made assumptions that underestimate, rather than overestimate, the rate of climate change, is the oceans’ ability to be a carbon sink. That has diminished considerably and will lead to higher atmospheric CO2 levels than assumed earlier.