Arnold Kling  

Predictability of Crises

PRINT
Ezra Klein Won't Take My Medic... Virtual Cornucopia...

Gilles Saint-Paul writes,


any macroeconomic theory that, in the midst of the housing bubble, would have predicted a financial crisis two years ahead with certainty would have triggered, by virtue of speculation, an immediate stock market crash and a spiral of de-leveraging and de-intermediation which would have depressed investment and consumption. In other words, the crisis would have happened immediately, not in two years, thus invalidating the theory. Thus, most crises are by nature unforecastable. Believing that they should be forecast is actually a positivist fallacy based on a false analogy between economics and the physical world.

David K. Levine writes,

our models don't just fail to predict the timing of financial crises - they say that we cannot. Do you believe that it could be widely believed that the stock market will drop by 10% next week? If I believed that I'd sell like mad, and I expect that you would as well. Of course as we all sold and the price dropped, everyone else would ask around and when they started to believe the stock market will drop by 10% next week - why it would drop by 10% right now.

Another argument they could make is that if policymakers saw a crisis coming, then they would take steps to stop it, so that it would not happen. Thus, any crisis that does occur has to be one that was not forecast.

But be careful here. I think these are good arguments for why a generally-believed forecast for a crisis can not be correct. But there is plenty of room for a given individual or method to forecast a crisis and not have enough people believe that forecast to take actions that would prevent it.

Thanks to Mark Thoma for the first pointer, and also for trying to keep track of the various points of view in the What's Wrong with Macro? debate.

In the end, it will be interesting to compare the quality of the debate that takes place in the blogs with that in academic journals. In the journals, there may not be enough confrontation between opposing points of view. In the blogs, there is plenty of confrontation, but far too much vitriol for my taste. Paul Krugman, whose New York Times piece provoked much of the recent blog controversy, is not in the habit of bending over backwards to put his opponents in a good light. I think that other economists would do better not to adopt a such pejorative style, even if they agree with him on substance. Overall, a lay person would get the impression that macroeconomists are like cable TV partisans--all antagonism with no substance. A cynic might say that this is in fact the true state of macro, but even I don't think it's that bad.


Comments and Sharing


CATEGORIES: Macroeconomics



COMMENTS (17 to date)
Urstoff writes:

Arnold, apropos of this post, what do you think of this take on recent macro by philosopher of econ Alex Rosenberg: http://leiterreports.typepad.com/blog/2009/09/alex-rosenberg-on-cochrane-and-economics.html

The Titanic:
-- Built of substandard steel plate.
-- Anti-flood compartments with separating walls
  that did not go all the way to the compartment ceilings.
-- Only enough life boats to meet outdated regulations, not
  to serve the actual passenger population.
-- A captain who maintained an unsafe speed.

No "macro-marine'nomics" could predict the Titanic disaster, but it would not have killed 1,490 people if any one of these had been done right.

Disasters are possible when people skimp on engineering in complex systems. No theory is going to predict how that disaster will play out, or when.

The U.S. economy is manipulated constantly by government and the Fed. No theory is going to predict when and how badly the economy breaks under these many weakenings.

More amazing, many people think that "injecting printed money into the economy to provide liquidity" is going to fix things. Another name for that is "pretending that there are more resources available than there really are".

If government spending produced wealth then we would all be in lounge chairs in Aruba by now.

Where is the macro-economic theory that predicts what happens when you promise to raise taxes in the near future on the most productive people? Amazing. They act like those taxes have been raised immediately. I would call that a corollary to David K. Levine's suggestion.

Jody writes:

think these are good arguments for why a generally-believed forecast for a crisis can not be correct.

Seems like a problem in need of a good fixed point theorem.

Bill writes:

predicting rain on monday doesn't cause meteorologists to carry their umbrellas to the car on friday night... Or even monday morning, in absence of rainclouds.
if your economic models don't realize this, then your economic models suck. that's the whole point.
1) perception of risk influences speculation as much as belief in a predicted outcome.
prediction does not mean certainty. maybe you should plug that into your model.
2) dispersal of information is what allows party A to profit from party B in markets, or do you believe that goldman sachs' meteorologists are handing umbrellas out on wall st to make sure no one gets wet? & assuring efficient markets for us (& profits for none)? is that what you clowns believe in?
or do your FED-sponsored models serve the same purpose as goldman's sell side research (while they hold huddle meetings with their real clients), shorting the same trash mortgage securities theyre hawking to their 'dumb money' brokerage customers. lol.
does anyone besides the 'dumb money' of the economics professsion actually swallow the idea that bernanke, et al. actually believe the 'efficient markets' garbage they're selling to the rest of economics?-on behalf of their real clients-the banks? who they provide with information to trade on, and funds to trade with? and professionally manage information flow to influence markets...to benefit some parties at the expense of others? while putting out reams of data to the effect that this very thing "can't happen?"
according to their very best models, of course.
is this what's happenning?

William Apel writes:

Everyone can explain why the dollar rose or fell yesterday. No one can predict what it will do tomorrow.

Don the libertarian Democrat writes:

I still read mainly econ blogs because, in my view, the quality of the bloggers and commenters is of a higher quality than political blogs, say. But I couldn't agree more with you about the vitriol. I can still find value by ignoring the caustic comments, but I don't think that it aids understanding and learning to be so personal.

Actually, Anthony Giddens has written a lot about the issue of the interaction between theories of human agency and the influence of such theories on human agency. See, for example, this page:

http://en.wikipedia.org/wiki/Anthony_Giddens

"Thus, there is a "Duality of structure", according to Giddens. With that he means that social practice, which is the principal unit of investigation, has both a structural and an agency-component: The structural environment constrains individual behaviour, but also makes it possible. He also notes the existence of a specific form of a social cycle: once sociological concepts are formed, they filter back into everyday world and change the way people think. Because social actors are reflexive and monitor the ongoing flow of activities and structural conditions, they adapt their actions to their evolving understandings. As a result, social scientific knowledge of society will actually change human activities. Giddens calls this two-tiered, interpretive and dialectical relationship between social scientific knowledge and human practices the "double hermeneutic".

Troy Camplin writes:

It seems to me that there is a difference between making a specific prediction about a specific situation to any degree of accuracy (the bubble will burst in 2 years, for example), and having a model that predicts that when certain things happen, you get a bubble, and that the created bubble will then burst under x conditions.

This is precisely the problem with how mathematics is used in economics. Mathematics is good for creating general models that help you to understand more or less what is going on -- but it is utterly useless to understand in any precise way what is going on. Mathematics is only a precise approximation of reality -- it is not and can never be a precise duplication of reality. This is how, I believe, mathematics has ruined economics, including macro, and contributed in no small part to the current depression we're in. Positivistic scientism could not describe economics prior to the Great Depression, and positivistic mathematicism could not describe economics prior to this depression.

The use of mathematical models to help us to understand how the economy works in general terms is great -- but we have to stop believing that these models in fact describe reality in all its messy complexity. That is the problem with macro.

Kurbla writes:

Theoretically, effects of the predictions can be calculated in, including the rate of the acceptance of the prediction among general public as a variable that depends on education, culture of the population.

Dont give up too easily.

woupiestek writes:

Saint-Paul is wrong on more accounts: a forecaster that only makes selffulfilling profecies could be right all the time.

Bill writes:

"...any macroeconomic theory that...would have predicted a financial crisis two years ahead with certainty would have triggered, by virtue of speculation, an immediate stock market crash..."

-what model claims anything with certainty?

-what speculator only bets on certainty?

-why would someone who had *certainty* of a crash and was speculating on it share that priceless trading edge with other market participants?

-wouldn't he place his bets quietly and wait for his expected outcome and maximize profits?

-isn't that precisely what multitudes of traders, hedge fund managers, and whole firms did?
(john paulson, peter schiff, goldman, etc.)

“No One Saw This Coming”: Understanding Financial Crisis Through Accounting Models(2009).

http://mpra.ub.uni-muenchen.de/15892/

Bill writes:

"...the crisis would have happened immediately, not in two years, thus invalidating the theory. Thus, most crises are by nature unforecastable."

-what 'crisis'? the 'crisis' aspect of this sideshow didn't begin until paulson and bernanke decided to play good cop/bad cop with a room full of congressional leaders, threatening martial law if the TARP vote didn't go through...

not until they NEEDED a public panic did they start throwing around the 'D' word and screaming about a 'crisis,' if they didn't get their 700B.

and not until their lehman switch and bait did they really start to freak everyone out, which worked out well for goldman sachs and morgan stanely...

next, they stuffed bank of america -- one of the biggest, fdic-backed deposit banks -- with the insolvent remains of merril lynch at something like 3x market price, literally a day away from crashing and burning. This on top of BAC's acquisition of Countrywide's toxic waste.

bear and wamu were handed to jp morgan minus their losses.

wachovia, w/ it's goldenwest shitpile was snatched from citi and given to wells... apparently the fed realized citi couldn't be stuck with wachovia's losses without dropping dead on the spot.

PNC got to swallow national city's trash heap, and finally, the fed and fdic took on goldman and morgan's risk by converting them to bank holding companies, thus availing them discount window financing, customer deposits, deposit insurance, and then letting goldman trade for a year with VaR metrics off the charts, still unadjusted to it's bank holding co. status...

to bailout their systemically insignificant buddies on wall st. they deliberately infected the entire commercial banking sector with the concentrated losses of countryide, bear, merrill, morgan, goldman, wamu, wachovia, goldenwest, national city, etc.

in some cases damaging institutions (BAC, LEH, BSC), in some cases granting them windfalls (JPM, GS), in all cases weakening, threatening and concentrating risk and loss in the largest, most systemically significant commercial-deposit banks, exposing their capital to the consolidated garbage of the entire shadow banking system and the worst regional/peripheral banks and mortgage brokers.

they took isolated, concentrated pools of risk in small and peripheral institutions, and infected the largest, most systemically exposed banks (with the majority of the country's deposits, locations, commercial credit, and more than 80% of the total derivatives market) -- the fdic backed, FED regulated deposit banks! -- with the debts and the losses of wall street, fraudulent mortgage lenders, reckless thrifts and regionals.

in order to shield the rich creditors and shareholders of these banks from loss, in order to socialize the losses of the frauds and failures and shield them and their assetts from the bankruptcy courts, from price discovery, from fair market writedowns.

in order to pick winners and losers and extend and consolidate control over the whole banking system and its maintain the value of its assets -- keeping the public in hock to failed banks, bailed out with their own extorted tax money!

and allowed to continue to collect on and profit, risk-free, from assets that should and would have been sold in bankruptcy at pennies on the dollar to solvent, prudent competitors, investors, private-placements, funds, etc., and IMMEDIATELY written down to a level slightly above the market-clearing price... thus making the loan both servicable and profitable at the same time.

Mike writes:

On prediction:

Nobody expects economists to precisely predict a crisis. Economists are being blamed not because they failed to predict the crisis but because they ruled it out. Macroeconomic models are being blamed because they did not allow for a crisis. Economists are being blamed for claiming robustness when, in fact, there was fragility.

fundamentalist writes:

So this is what mainstream macro has come to, everything is random and therefore unpredictable? Hayek wrote in his Nobel acceptance speech that we study economics because we want to predict the future to some degree. Now mainstream economists are saying that is futile. So there is no reason to study economics?

Mainstream econ cannot predict crises only because we do not understand their causes. And if mainstream econ doesn’t understand the causes, it also doesn’t know what prescriptions to take to fix the economy after the crises start. A medical doctor would be guilty of malpractice if he prescribed a treatment for an illness he didn’t understand. Mainstream economists are guilty of malpractice when they recommend monetary pumping as the cure for all economic disorders, even though they admit that they do not understand the underlying malady.

That crises are random events means nothing more than that the causes are two numerous and their interactions too complex for us to grasp. If that is true, then how does it follow that a single solution, monetary pumping, is always the cure for every crisis?

Also, if the above is true, then economics simply cannot be a science because no general principles of how economies work are possible. In that case, the past century of economics has been a complete waste of time. We should resurrect the old German Historical School and the old Institutional School and admit that economics is nothing but a sub discipline of history. All we can hope to accomplish is to catalogue the specific events of every crisis in as much detail as possible and never, under any circumstances, try to see patterns in them because according to mainstream economics it’s utterly impossible for patterns to exist, because economic principles are nothing but patterns in the data and in human behavior. And if any patterns existed, then some form of forecasting would be possible, too.

BTW, it’s always interesting to me that mainstream economists accomplishing the impossible the requirement. If you can’t forecast with 100% accuracy the value of the S&P 500 at 11:15 AM on Wednesday, then any kind of forecasting is impossible and completely worthless. Anyone who has studied the stock market for a few minutes knows that predicting the price of stocks on a weekly basis is impossible. Predicting monthly averages is a little easier, while quarterly and annual averages can be fairly good.

I would like to know of any mainstream economist who did not lose money in the latest bear market. I know of a few Austrians who did not. I’ll bet that mainstream economists waited until the market hit bottom to sell their index funds and they haven’t entered the market again, so they have missed the 60% run up in stocks since then. People who make money in the stock market take it from mainstream economists.

Niccolo writes:

Perhaps it's not been stated yet, but there are also vested interests for not moving on a recession early and there are interests to ignore an otherwise impeding doom to the economy.


Many economists still do not seem to take into consideration any public choice theories when they make these statements.

bgc writes:

(This may also be what Niccolo is saying)

I would say that it depends on incentives - and the weighting of short vs long term. In bubbles there is a situations in which it only makes sense to be short termist, because long termism is punished

eg in the middle of a bubble a mortgage manager who ensured that only good credit risk clients would get a loan would find his section made less profit than a manager who allowed the sales staff to loan to everyone. And a careful saleman (on commission) makes a lot less than a salesman who will sign up anyone.

A bubble can certainly happen even tho' everyone knows it is a bubble - because the only people who lose are those around when the bubble bursts.

Until that point the participants will simply be trying to keep the bubble growing and try to exit at the last minute before the burst.

In fact isn't that exactly what happened?

The only thing to do when you know you have a bubble ongoing is to make it burst immediately - the longer the delay the worse the damage.

Only government can do this - none of the participants can do it, because those who behave long termistly will destroy themselves.

But of couse governments instead fuelled the bubble and kept it going an extra three or so years.

So the key piece of information needed is not the timing of a down-turn or bust, but the simple fact that there is an ongoing bubble, with the characteristic _behaviour_ of bubbles - such that short-termism is being exclusively rewarded and long termism being punished.

It is the timescale of motivation that defines the bubble as such - nothing else.

Jeremy, Alabama writes:

Excellent post.

But does deliberate crisis management create results more like:

a) Flu epidemic: deliberate management slows spread, provides time for effective treatment, and prevents overwhelming emergency services

or b) Forest fires: deliberate management extinguishes small "preventative" fires which leads to gigantic out-of-control wildfires.

In other words, we can't even answer the question - is management of economic crises helpful or not?

a student writes:

we hear different statements about crashers and the crisis by different economists. namely

1- crashes are not predictable
2- Not much can be done to avoid crashes or lessen its severiety
3- Too much reliance on EMH and rational expectations (or mathematical beauty) has pushed economists not to study crashes as they should
4- Too much reliance on EMH and rational expectations leaded to excessive deregulations which in turn made a such crisis to occur
5- Some regulations has made the crisis more severe.

Can anybody evaluate all these statements in one place?


Comments for this entry have been closed
Return to top