Arnold Kling  

Certainty, Uncertainty, and Macro

Thoughts on Probability and Un... Take That, Math Geeks...

Steven Randy Waldman writes,

A housing boom, any kind of boom, is attended by an increase in certainty. Information is stimulus, confusion is contraction. A bust occurs when the market is unsure of everything, when market participants perceive better risk-adjusted return in holding government securities (or supply-inelastic commodities) than in financing real investment. Sectoral shifts per se have no clear implication with respect to variables like employment and output. But "hangovers" do happen, because powerful booms are periods when market participants make consequential decisions with great swagger and confidence, and busts are when we learn that despite their certainty, they were wrong. They are left not only impoverished and burdened by debt, but bereft of confidence in their ability to evaluate new opportunities.

Comments and Sharing

CATEGORIES: Macroeconomics

COMMENTS (11 to date)
david writes:

This is a very strong assertion. Is there any empirical evidence? It implies (say) that the few market participants who correctly judged the end of a bubble would now be correspondingly confident of their ability to evaluate investment opportunities during the following boom. There were a few such people, why aren't they marching triumphantly down Wall St.?

In any case, this is becoming suspiciously close to animal spirits and hydraulic macro. Is certainty truly all that is at work? If the government suddenly stands up and says, okay, now we're nationalizing this and this and this troubled industry and fixing relevant prices, it's all very certain (if certainly bad, certain nonetheless) - does this rapidly restore employment and growth in other sectors? Doesn't seem very plausible. What happened to the intersectoral calculation at the base of the model?

Vasile writes:

As of last (half of year?) I've started thinking of markets as of producers of prices, not goods.

So both the Arnold's "Recalculation" story and Steve Waldmann's "certainty bubbles" story resonates rather very well with me.

But if you think of markets, as of finite state automata, whose main (and only) input is AD, and main (and only) output is GDP... well, I don't know and will not guess.

Steve Waldman writes:


there're plenty of investors who did okay through the bust and continue to invest with confidence, perhaps with more confidence than before. but in dollar-weighted terms, they are a very small minority, so in aggregate we see a reduced capacity for certainty, or greater resistance to informing the market in the sense described below. investors, especially successful investors, rarely hold parades.

i use information in a particular technical sense that i think leads to testable propositions. the market is more informed when the entropy of investor behavior is reduced. i claim that information, in that sense, is coincident with booms. subdivide history into time periods (granularity TBD, too short we can't measure entropy and catch idiosyncratic news-cycle artifacts, too long we span regimes). operationalize capital flows (ie investor behavior) as stock price changes. that's terribly imperfect -- it wouldn't pick up cap flows into structured finance for example -- but it's easy. divide the stick market into sectors, by SIC codes or somesuch, and build portfolios. for each subperiod, find the empirical distribution of the (vector-valued) sector-price-change random variable (adding back dividends), and measure its entropy (you probably have to approximate the observed distributions as joint normal or something, but you'd do so consistently across time periods). i claim that during boom periods you find a lower entropy empirical distribution than during busts or ordinary periods. (there're lots of issues -- e.g. i would want to excude consideration of flow to gold or treasuries, and limit my claim to the entropy of "risk-asset" capital flows.)

if you read the original piece, i hope i make it pretty clear that certainty is not necessarily good. i think that arbitrary certainty created by government fiat is roughly equivalent to an asset price bubble (unless the government happens to be wise or lucky in its choices, which i do think is possible, see Bryan Caplan on dictators). imposing artificial certainty would indeed serve as stimulus, create a boom that would increase measured outcome and employment. but absent luck or wisdom, the stimulus would push the economy in a direction that is ultimately unsustainable, and there would be a bust, just like with any other bubble.

by the way, i think a better market wouldn't have the characteristic i describe (and propose looking for) above. in a better market, investor information would be more fine-grained. certainty and overall capital flows would still coincide, but the views about which investors are certain wouldn't be so aligned that you'd observe information in sectoral-level flows. my conjecture above depends on certainty borne of herding or unreliable information cascades that would be mitigated in more intelligently organized markets.

Vasile writes:

david: It implies (say) that the few market participants who correctly judged the end of a bubble would now be correspondingly confident of their ability to evaluate investment opportunities during the following boom.

I don't see that implication. The situation is not symmetric. Take Peter Schiff, he was famously right about the housing bubble, but less so about the following events. Things like this can cool one's confidence rather rapidly. The more activist than usual policies of governments and central banks do not help either.

And even assuming for the sake of argument, that you are right, you'll have a confident minority vs. a bamboozled majority. Well gold traders have the time of their lives, but this is still a marginal issue.

And even supposing (again) that the gold is the wave of the future, and all have become confident about this, the gold extracting industry will not be able to "suck in" all currently unemployed workers over a quarter or so. Capital is not that homogeneous. Infrastructure will have to be expanded, equipment will have to be produced, which is different from the one which was produced during the housing boom and is now idle.

And no, govt. intervention will not help, but this is already a long post.

Lee Kelly writes:

Though I hardly think it the whole story, I have been saying this for months.

When the housing bubble burst, a lot of people realised they didn't know what they were doing. It was as though the laws of physics just changed in the course of a few weeks, and, understandably, people lost confidence. But what they really lost was knowledge, because the boom sustained a delusion by distorting relative prices.

Sometimes schizophrenics become attached to their delusions, because in their fantasies they are important, capable, and interesting people. In real life, meanwhile, they are nobody of consequence, and sometimes it is more satisfying to return to the fantasy.

I see the government as encouraging a relapse, and many in the market are awfully tempted to return to the pleasent delusion.

ThomasL writes:

I think this has some truth in it.

That uncertainty has a cost I don't imagine anyone would deny. The reverse is that certainty lowers the cost; and the more certain something is the cheaper it becomes. Bubbles develop a pretty strong "certainty effect", drawing investment into the "sure thing" until it crosses into a level of certainty that is almost a forgone conclusion: "You can always make money in X."

I'm not sure it is calculable, but if a dollar value could be attached to that risk-numbing "certainty effect" it would be interesting to see.

I'm not sure if I agree with causation, however. It wouldn't necessarily follow that the bust occurs _because_ of uncertainty creeping in. That has a kind of stilt to it that sounds as if it is ignoring the possibility that something truly could be grievously wrong. It is the standard topsy-turvy idea of the bust being the problem, not the mistaken confidence/certainty that preceded it.

fundamentalist writes:

"A housing boom, any kind of boom, is attended by an increase in certainty. Information is stimulus, confusion is contraction."

The second sentence doesn't logically follow from the first. The fact that certainty "attended" the boom does not mean it caused or stimulated the boom. Neither does confusion cause the contraction, but it "attends" the contraction.

Business failures due to capital goods shortages cause the switch from certainty to confusion. Wide-spread failure results from the distortion of prices caused by the Feds' loose monetary policies. Businessmen proceed with confidence that the distorted prices are actually correct, then confusion sets in when their plans fail.

fundamentalist writes:

Eric Falkenstein on the previous article: "What is needed is something constructive, something the Austrians, Post-Keynesians, or Taleb, have failed to do."

This statement ties into this article in the sense that failure of predictions causes confusion. That failure is due mainly to a faulty monetary theory on the part of mainstream econ. But it's also due to a lack of the necessary data. As Hayek said in his Nobel speech, we will probably never have the necessary data. so the kind of certainty that Falkenstein is looking for probably is impossible.

floccina writes:

This goes along with my belief that what people call sure things are he riskiest investments and that good investors lack self confidence.

Also people though risk averse, take great risk due to lack of knowledge.

If you think of information in terms of Claude Shannon's definition of it, then you could say that information is extensive during periods of certainty and more correlated (hence non-extensive) during periods of uncertainty.

If there is then a link between the volume of information space and the flow of investment, then you might see the shift between extensive and non-extensive information space reflected in market conditions.

Walt French writes:
A bust occurs when the market is unsure of everything…

Certainly the VIX is a very good measure of stock market investors’ “uncertainty” and yet it lagged the pronouncement of the NBER which itself was quite a bit after the peak of employment and when home prices, foreclosure stats, etc., were getting more than a bit wobbly.

The VIX, stated in standard deviation terms, is backed out of options traded by a wide swath of the market; if it is too pessimistic (high volatility) then you can make lots of money selling people options that are very unlikely to be paid out (“expire in the money”), and if it's too low, you can profit from buying options that use that volatility, expecting fat returns when the market moves around more violently.

It's inconvenient to chase down the exact data, but my recollection is that the VIX -- market uncertainty -- peaked (at nearly 4 times its current value, at the time when stock prices were falling the most virulently, shortly after Lehman failed a year ago. The administration had already signalled no bailout for the illiquid, and probably insolvent, firm, so uncertainty was about how rapidly the disease would spread, not whether we were in a recession, nor whether the banks were in trouble.

This single observation (sorry I can't document it now), disagrees with the notion that late-to-the-party uncertainty drove down the market. And it's violently at odds with the notion that uncertainty drives money away from investment opportunities.

Meanwhile, Finance's reigning theory is that investors are more compensated for more systematic risk-taking, in order to induce prudent people to back startups like Google, where the odds are much tougher to calculate. I.e., the assertion flies in the face of theory, too.

SRW writes well enough, but it looks like rationalization, not theoretical or empirical work. Anybody quoting it should append, "these statements have not been subjected to minimal review by anybody."

Comments for this entry have been closed
Return to top