Krugman's Intro to Keynes's General Theory
Submitted by
CalculatedRisk on Tue, 03/07/2006 - 6:33pm (23 comments)
Shopping at Amazon?
Wait! If you click this link
Open predictions
Our local book authors
User login
Who's online
There are currently 17 members and 134 guests online.
Recent Recommendations
"Keynes is your enemy. And he is an especially dangerous enemy because his ideas have been vindicated so thoroughly by experience."
Apparently Krugman is our enemy too, as he tries to pull the wool over our eyes. "Vindicated thoroughly" is a subjective conclusion at best, and is in reality an outright falshood. All Keynes has shown is a way for governments to distort the free market and pretend it is all good economics.
Free market proponents do not claim the free market produce the best of all possible worlds. The only claim is that government is always worse in the long run.
A market is a feedback-controlled system. In control system engineering, the first thing one learns is that although with proper tuning feedback control can give stable operation, with improper tuning it will cause the system to break into oscillation. This is especially so of systems in which there are long lags and/or dead-times (pure delays) around the loop; to stabilize such systems one must strictly limit the gain (amplification ratio) around the loop. Basically, there's a limit to how responsive you can make a feedback-regulated system, and if you go beyond that limit by a little you get "ringing" after an upset, and if by a lot, wild and destructive oscillation.
As the free markets of which libertarians dream are free of all government regulation, their "tuning" (or lack thereof) is a matter of chance. The result is the boom-bust cycles we see in those markets with long lags and dead-times; the real estate market is a perfect example. The only way to prevent boom-bust oscillation is for someone to tune the gains, which essentially means imposing limits on responsiveness (e.g., imposing limits on leverage).
It's important to understand that the effects of lags and dead-times on feedback-regulated systems are mathematically deterministic. The only way to get more responsive control than the limits allow is to use approaches other than pure feedback control whose economic analogs would be equivalent, for example, to the Fed anticipating asset bubbles and proactively imposing restraints to prevent them (which, of course, they claim is utterly impossible).
jm - What a wonderful comment.
It misses of course the obvious first response that free market economics is itself an artificial construct whose justification is entirely its pragmatic effectiveness.
And of course he leaves himself wonderfully open to the classic Keynesian put down....
"... but in the long run we are all dead!".
But as confirmation, I worked for a while on a simple economic model built on the assumption that individual markets all adjusted towards towards desired tragetories over the long term. Despite this, eigenvector analysis indicated that the system was (slightly) unstable. This lead eventually to the abandonment of the model by the sponsor as it didn't meet his preconceptions. Brilliant!
And then there is the strong historical evidence that stabilising policy (or at least the growth of the public sector) has reduced the amplitude of fluctuations substantially in recent decades.
I'm a bit afraid that the current prevalence of neoliberalism and the resultant financial deregulation is allowing the danger of instability once more to grow. All the defences we built in response to the Great Depression are being pulled down!
the really incredible thing is that there still is an austrian school of economics !
jm-
Of course those who are doctrine-diven true believers in free markets don't just dislike active government intervention. They also dislike structural restraints on markets designed to "detune" markets and prevent "excursions." Things like automatic trading limits on the stock exchange are seen as restraints on a free market and are a priori, "a bad thing."
The thin deck of the Tacoma Narrows Bridge meant that the STATIC wind loads on it were lessened, and it was a more "efficient" design. However that design ensured the bridges self destruction in a wind driven dynamic oscilation. Extreeme freemarket theorists often remend me of the engineers who were didn't look at the dynamic response of the bridge to wind, instead relying on simpler, static wind load calculations.
It is a very difficult thing to try and figure out which market inefficencies just might be a good thing. Especially since technology improvements change natural damping and feedback response.
jm,
A wonderful explanation of optimal control theory. But those familiar with recent economic history would say that the applicability of optimal control theory, which was in vogue in the 1960s, to economics is limited by the "time inconsistency" critique and the "lucas critique" (the respective authors got Nobel Prizes for this finding).
Unlike a radio or any physical system, the economy is populated by "actors" who are self-aware and can "learn" about the "system" they inhabit. If anybody tweaks the system, then in due course, the actors would modify their behavior to the new realities, thus undermining the system. For example, if financial markets know that they are going to be bailed out, then there will be more risk-taking, making the system more unstable. We have seen all this happen in the Greenspan era.
However, while I think "fine-tuning" is indeed counterproductive, big government spending as a flywheel can be quite effective while avoiding the pitfalls of fine tuning. In short, I dont support the Greenspan kind of activist monetary policy, but I support the large automatic fiscal stabilizers. The difference is the latter are not "discretionary", while the former is. Unfortunately, most modern keynesians (I mean the new-keynesians) support activist monetary policy. The Greenspan era, the fallout of which will be in the years and decades ahead, will reveal the hollowness of this approach.
Jim A.: Exactly so. The Tacoma Narrows Bridge disaster is a very good example of the hazards of overly "efficient" engineering designs. Henry Petrowski's "Design Paradigms: Case Histories of Error and Judgment in Engineering" has many more.
When I read the writings of economists, I am repeatedly horrified by the degree to which so many of them blithely engage in policy-making activities which are for all practical purposes forms of engineering, with apparently complete ignorance of the history of engineering.
You are correct that it is difficult to figure out which market inefficiencies will be good, as evolution of the economy changes the nature of its response. Also, as Bernanke comments in one of his books, conventional optimal control approaches aren't directly applicable to economic systems because they include intelligent agents whose behavior changes adaptively, and who in some cases will actively "game the system".
But I can't accept the current Fed position that asset market bubbles can't be anticipated and forestalled, and that all that can be done is to mop up afterwards. Just plain common-sense regulation of mortgage lending -- requiring reasonable down payments to prevent high-leverage speculation -- would have prevented the current housing bubble.
I consider today's speech by Poole of the Fed outrageous. Among his comments are, "... Over the five-year period, real prices increased by 61 percent in Boston, 112 percent in Los Angeles, 100 percent in Washington, D.C., and 26 percent in St. Louis. ... ", and "from the late 1980s to the mid 1990s, real house prices declined by 30 percent in Boston, by 36 percent in Los Angeles, by 20 percent in Washington, DC, but by only 11 percent in St. Louis". Yet he is clearly quite complacent about this, opining that there is no "nationwide" housing bubble, and speaking overall in a tone that I interpret as utterly complacent.
No more time to write this morning. Just one last comment -- how are the home prices increases Poole quoted consistent with the Fed's claims of low inflation? Yes, I know how CPI is calcualted using owner's-equivalent-rent, but the Fed knows dam well it's baloney.
2 points:
In my earlier post we should replace "designed to" with "having the effect of". Designed implies much more active control than I think actually occurs.
I don't think that owner's equivalent rent is bunk exactly. It is a simplification inteded to allow one to look only at the underlying market for housing, separating out the speculative yoyo. There's nothing wrong with that, unless you tries to use it to prove that speculation doesn't happen or isn't important. Just as with the bridge example, there's nothing WRONG with looking at the steady state wind loads. In the absense of any dynamic coupling it's still very important to make sure the a bridge doesnt blow over in a strong wind.
Another engineer pipe up...
I too took 'control theory'... I took both mechanical system dynamics & chemical engineering control systems... The former focused more on building stable systems that resisted the instabilities caused by 'disturbances'... the later fully recognized it was impossible to build a fully static steady state chemical process plant so you need a LOT of carefully crafted active controls.
They both work in the right applications & they both fail in the WRONG application... the Tacoma bridge is just one very visible example... I was in meetings in Iowa all day yesterday addressing a much small & less visible case of the same kind of problem.
As it applies to economics... something I know DAMNED LITTLE about and so feel completely at ease discussing...
I think the Austrian' approach assumes free market systems are inherently 'more stable' due to the 'natural feed back' jm mentioned. The perfect market myth.
I believe the Keynesians are in the camp that disturbances happen (think of the bumper sticker on a similar theme) and as a result SOME control & damping is necessary or at least desirable.
I don't see anyway you reconcile these two beliefs because it is VIRTUALLY impossible to test & verify in a laboratory like you can physical systems. So they will both forever be to a certain degree 'faith based' theories.
I believe you can get an approximation for the 'truth' through computer modeling... but it will never change the minds of the true believers in either camp... they will base any results that contradict their view on 'GIGO' garbage in garbage out.
More to follow
As for tea's point that perturbing the system via 'control' doesn't work in economics because the players 'game' the system... is only half true... if the control system anticipates some of the 'gaming' it can still work very well (usually via a control process we call 'feed forward control)... It is harder to do but possible. Couple a partially successful 'feed forward strategy' with an appropriately damped & sensitive feed back control system and even the 'smartest' gamers can be kept in their corrals...
I learned feed-forward-feed-back control running 'continuous fermenters' at a large ethanol plant... those 'bugs' aren't 'stupid'... if given half a chance would 'take off' and ferment their way right out of the tanks like shaking a can of beer before popping the top... only our cans were 100,000 gallons each. Feedback just wasn't enough... so we anticipated the future response to any step ups or spikes as they happened (based on models & experience) then let the feed back control systems dampen out the remaining fluctuations after having been given the jump start. It was in effect an 'Expert System'... almost 'AI' like.
The problem as it applies to econ is that econ is more complex & the 'experts' to a great degree don't understand what is going on. Think of econ as as an organic system where the process evolves in tandem with the control system.
In general it is difficult to employ feed back if you don't know how the system responds to controls. It is impossible to employ feed forward without excellent understanding of how the system responds to controls... That is the Achilles heel to applying aggressive control theory to econ.
However I believe there are ways to get a grip on the basic information needed to apply 'controls'. As I hinted in my first comment I think a lot could be learned by computer modeling matched to empirical measurements in the economy. The thing I would do differently would be to build up the models from 'micro-econ principles and not deconstruct from top down from macro principles.
The analogy to physical systems could be seen in either statistical mechanics (as in P Chem) where large complex chemical processes are very accurately predicted by looking at individual molecular interactions even down to the quantum level then summed up as a whole macro system.
Another example would be Finite Element Analysis (either mechanical or thermal) where systems (typically parts) are mathematically broken up into little pieces (grids) and then the whole grid is modeled to see how stimuli (typically loads or forces) in one place pushes through the whole system.
In short there are ways to anticipate the gaming both in advance & as it evolves but it takes a lot of information & excellent metrics (boundary conditions).
And just as past performance is no guarantee of future returns, many systems will exhibit linear feedback until they don't. Pulling back on the stick make the plane go higher...until it stalls. I harbor a suspicion that the current U.S. economy is so debt-choked that the previous macroeconomic tools that the Gov has used in the past are loosing their effectiveness.
tea, Jim A., dryfly, reason:
Wonderful comments!!
(too bad I'd started my 10:40 before tea sent his 9:58 -- would have referenced its rich content rather than just skimming over the issue)
dryfly, your batch fermenter post is a classic. I will treasure it.
dryfly, your batch fermenter post is a classic. I will treasure it.
jm they were a series of stirred CONTINUOUS fermenters! Yeast & broth (beer mash) in one end & partially fermented 'beer' out the other... which then fed the next fermenter. These were big tanks - each about 100,000 gallons. We had a whole battery of them and they worked great most of the time.
Fermenting yeast produces a lot of heat... so we had cooling coils in the tanks to pull this heat away. The cooling water was circulated through evaporative cooling towers outside. No problems in Fall, Winter, Spring... but on hot humid Iowa summers - especially nights when the outside relative humidity approached 100%... BOING - those tanks would erupt.
The yeast would get warm... the warmer it got the more they fermented... the more they fermented the more gas they generated (CO2)... and eventually they would blow their top... 3-4 tanks would simultaneously eject a stream of partially fermented 'beer' 10-15 feet into the air out 3 foot dia top 'manholes' like Old Faithful... covering the grounds with a sticky sweet alcoholic slime... including the workers cars. One night we estimated we blew 180,000 gallons out the top & about half of it ran into the river as it overtopped our containment. There was nothing we could do.
As a young chemical engineer back then I learned plenty about 'unstable systems'. Good thing we weren't running nuclear piles.
The original Keynes free-for-all over at DeLong's parallels this in an cool way. The new & most fabulous thing in economics is evidently Dynamic Stochastic General Equilibrium (DSGE) models. Based on difference equations rather than your engineer's differential equations, ew, with endogenous and exogenous variables nicely segregated in separate vectors per control theory. Autoregressive models serve to test the formulation, and the models give you the same kind of conditions for damped and explosive oscillation.
The new & most fabulous thing in economics is evidently Dynamic Stochastic General Equilibrium (DSGE) models. Based on difference equations rather than your engineer's differential equations, ew, with endogenous and exogenous variables nicely segregated in separate vectors per control theory.
psh - If my limited understanding is correct, the trouble with all of the hot econ models - including the old diff eq based control theory based ones is they are all 'top down deconstructive'.
That is they develop a bunch of equations of the 'macro system' based on macro level variables alone than use them to predict outcomes. All of this highly aggregated & homogenized.
They need to turn the problem upside down and 'construct' macro models based on micro level summations of mixed populations of 'individuals'. You don't need to know every person or firms behavior... just somewhat accurate 'behavior profiles' of 'communities' or 'populations' of individuals & also their distribution... and overall system 'conditions'... and lastly how these individuals in each seperate population are statistically LIKELY to behave given those conditions...
Based on that sum up the all the micro level behaviors & activities and predict macro level outcomes similar to how statistical mechanics in physical chemistry predicts rates of reaction & equilibrium mix & heat generated in reactive mistures based on individual molecular structure, the mix of molecules, their distribution and other variables like diffusivity and then predict dynamic system conditions (generally temperature & pressure & composition) that result.
Its a lot to know & calculate but its a lot easier to poll & then measure individual people & firms then it is to 'poll' molecules.
FEA (Finite Element) is done in a similar fashion for mechanical systems... though generally not as complex as stat mech in chemistry. Gives very good predicitve results once the 'system' is properly 'meshed'.
It was once incredibly difficult to do things like this prior to computers - but now with cheap computing power it should not be difficult... and if done right VERY powerfully predictive.
Somebody HAS to be doing this... if not some economist a whole lot younger & smarter than me should take a few P Chem or FEA classes & see if they can't stretch the models to cover 'human behavior' patterns instead of chemical reactions & mechanical stress strain relationships.
Just a hint.
I'm sure the Chinese are working on it Dryfly.
Krugman is a full-fledged Kook:
dryfly:
I, too, have been thinking along the lines of the ideas in your 6:14 post -- so of course I think they're excellent.
One possible limitation on how far ahead predictions could be made based on such techniques would be the "butterfly effect" that limits how far out you can do accurate weather prediction, but even so it would probably be better than the systems of linear partial differential equations I've seen in economics journals (that was decades ago -- maybe my criticism is out of date).
Jim A.: I somewhat agree with your 11:37 comments on owner-equivalent-rent. It has some utility. But there's something deeply wrong with the Fed complacency over the real estate bubble, and that metric has helped the Fed to avoid doing anything about it. I think we're headed for the worst real estate crash since the Great Depression, and that it will cause enormous pain to tens of millions of people, even to the degree of causing a significant number of premature deaths and physical injuries through its side effects.
I'd have to agree these are some smart comments from smart people. Unfortunately, the bias is always the same - we smart people can surely do better than the mania of the market left to its own devices.
Herein lies dead the myth of the smart central planner/bureaucrat/banker - take your pick. Stick a fork in it. The destruction of the markets may well be visited on us ten-fold, on account of the moral hazard, derivatives and other favorites of our central bankers. The market manias live on and grow, only they're not as obvious to the man on the street.
The one thing I don't understand is whether this is ever going to be obvious to everyone. Like a case of one's inheritence getting drained out by your trustee. If you did not know how much you really had, maybe you'd notice only when you're thrown out on the street.
the bias is always the same - we smart people can surely do better than the mania of the market left to its own devices.
that's not a bias, that's a moral standard !!!
you are completely perverse in calling this a bias.
smart and moral people must not that 's a moral requirement, go on with the crowd.
If there's only one door open and the crowd tries to run away from the burning theater, so hard that it prevents the opening of the second door, smart and wise people must calm the crowd open the door, have people evacuate in a serene order.
That's the same with housing and asset bubbles. visit The Center for Economic and Policy Research - CEPR
since 2002 the housing bubble is apparent.
Has something been done about it ?
No.
Could something have been done about it on the part of the ruling bodies (predidency, congress, FED) of course, they could have regulated the credit industry, they could have raised rates, they could have encourage renting with some schemes, they could have taxed capital gains etc etc.
they didn't do anything because they enjoyed the ride.
That's immoral. Because now all that's a ahead is a crash.
What have they done instead, they have adopted your so called bias less attitude ... Heck, we can't outsmart the market. Bogus.
Krugman is a full-fledged Kook
Observer... the game isn't over yet. I am not convinced 9/11 is as big a threat to our democracy & way of life IN THE LONG RUN as is the centralization & concentration of corporate power of which Enron is one symptom. I know of no republic or democracy that survived once an entrenched & nearly hereditary aristocracy became entrenched. I fear we are close if not past that mark.
This one might take a generation or more to know the outcome. I'll cut PK some slack until I can look way back.
Spectator - I live in 'markets' as person woking as a middle man in mfg supply chain. I think markets work great at some things... but definitely require good 'cops' else the powerful in the market set the 'rules' and manipulate for their advantage only. I see it every day... I'm about to lose a $5MM dolar contract in such a case. There is nothing I can do - no recourse, no appeal. I will lose money & about 10 families will lose a bread earner. It just sucks.
There is a sweet spot for sure... over-regulation & too heavy a control is as destructive as no controls, no rules, no regulation... The more complex & inter-connected the market the harder it is to find the sweet spot (just like in a massive chemical plant control process - thats why they 'monitor' plant wide but control locally as much as possible in decoupled pockets or cells via 'distributive control' strategies).
I'd feel a lot better about 'less market regulation' if we had better safety nets. In other words I'd be more willing to let some or even a lot of milk spill as long as there were plenty of mops ready before hand. Im not saying remove moral hazard by making individuals who took risk 'whole' but to make sure people don't starve & take to the streets as in the French Revolution or St Petersburg & Moskow in October 1917.
That in my opinion is the Achilles Heel of the 'Austrian Model'... They fortified themselves against too much gov't intervention & the resultant problems but left themselves completely open to the opposite extreme... and the political consequences can be every bit as dangerous.
"that's not a bias, that's a moral standard!!!"
Here's the problem - whose moral standard do you want to apply?
"since 2002 the housing bubble is apparent.
Has something been done about it ?"
Do you say nothing has been done about this by our smart central bankers?
On the contrary, they have encouraged and cheered it at every turn. The market without the promise of helicopter money would have corrected a long time ago. So much for applying the moral standard - which in this case included pandering to the financial community.
Yes, there are times when regulation is productive. If only everyone realized that less is always more with these things. Only then would it be possible to get out of the holes regulations dig for us.