Arnold Kling  

Bubbles vs. Non-Bubbles

Oil: Where Krugman and I Disa... Ignorant Masses, Dogmatic Elit...

From an interesting Wall Street Journal item:

Bubbles emerge at times when investors profoundly disagree about the significance of a big economic development, such as the birth of the Internet. Because it's so much harder to bet on prices going down than up, the bullish investors dominate.

The article describes academic research at Princeton. It says,

They are building on work done by the late Hyman Minsky, whose once-ignored ideas about investing manias are now in vogue, and the late economic historian Charles Kindleberger, whose 1978 "Manias, Panics and Crashes" is a classic. But compared with Mr. Minsky or another student of bubbles, Yale's Robert Shiller, the Princeton trio focuses less on mass psychology than on mathematical models. These they use to show how bubbles can be created even in markets that include rational, calculating investors.

A couple of comments.

1. Mordecai Kurz was doing this years ago, but no one seems to want to give him credit.

2. As Kurz would tell you, there is always disagreement about asset values. Moreover, if institutional factors lead to more trading in a rising market than in a falling market, then high trading volume could be a sign of rising fundamentals, not necessarily a bubble. So I'm not sure how one detects a bubble.

3. Robert Shiller thought that there was irrational exuberance in the stock market back in 1996. My guess is that if you were to go back and check, if you had bought stocks then, you would be ahead of the game today, meaning that stocks have outperformed risk-free assets by an amount that would exceed a reasonable risk premium. To put it more strongly, if you had bought a very long-dated put option on stocks back then that could only have been exercised ten years later, that option would have been worthless. If bubbles are detectable, then such long-dated put options should be winners.

4. Paul Krugman's incorrect reasoning notwithstanding*, there could be a bubble in the oil market today. But it is not detectable.

*I've thought about this some more, and I've decided to take a stronger view against Krugman's claim that refiners' inventories of oil have to rise if there is a bubble. His model is just plain wrong.

First of all, one cannot ignore forward prices. A rational market for assets that can be carried over time requires a long-term equilibrium value in order to pin down the path of expected prices going forward. If Krugman were here, I would remind him of the role of the long-term equilibrium exchange rate in the Dornbusch model.

What a high forward price (or futures price) in the oil market means is that the market thinks that we are in danger of running out of oil. The market is telling consumers to use less oil and telling producers to leave oil in the ground. In Krugman's model, the consumers listen, but the producers just blunder ahead, not reducing oil production. For rational oil producers, that is not a plausible assumption (yes, I know, there are reasons to doubt the rationality of those who control most of the world's oil reserves, but Krugman and I are both ignoring that here).

Incidentally, the forward price does not have to be above the spot price in order to induce producers to cut production. The impetus to cut production can arise from producers either (a) perceiving that the high price is indicative of potential future shortgages or (b) simply realizing that with the higher price, current demand will be lower, and they should cut back production in line with the lower demand.

I believe that there are bubbles. I believe that a bubble is a theoretical possibility in the oil market, although I do not think that one is occurring at the moment. But I do not think that bubbles are sufficiently detectable to make me want to have the Fed try to speculate against them.

Comments and Sharing

COMMENTS (12 to date)
ed writes:

According to Yahoo Finance, if you had bought and held SPY (an S&P500 ETF) at the end of November 1996, and reinvested all dividends, your annual return between then and yesterday would have been 7.2% (not inflation adjusted). That sounds like a reasonable risk premium to me, if a bit below historical averages. Buying at the end of any month other than November would have meant even higher cumulative returns.

Price yesterday=142.60
Price at end of Nov 1996, "adjusted" for dividends=64.35


Josh writes:

I think one problem is how the money from disturbances gets allocated. If you think about the internet, it has probably added trillions of $ to the world economy (think of all the things you either couldn't do before the internet or that would've been far more costly). That money is real, but where is it reflected? If you see that money coming (c. 1998) and you think it will be reflected in the profits of Yahoo, you'll put your money into Yahoo. But for whatever reason (perhaps because most profits get competed away regardless of industry), the money does not go to Yahoo and your Yahoo money is lost.

But the point is that the internet itself was NOT a bubble - the value of the stock market in 2000 was probably significantly less than the wealth the internet will actually create. The bubble was assuming that the money would be allocated to a few companies with .com in their name.

I think it's the same problem with oil - there is a disturbance now with energy (there's not enough of it vs how much we wish there was). How will the financial fallout from the disturbance be allocated? Some people are betting on it being in the price of oil - and there's a good chance they're right. But it could come in the form of some new energy source (not that I think it will happen, though), in which case the value of oil will plummet. Or there's probably a bunch of other things that could happen.

I think my basic point is that trying to predict a bubble is the same as trying to predict how the wealth the internet created will be allocated or what forms of energy will exist in 2015. It's not something you can get from a model.

Lord writes:

How does a lower forward price indicate a future shortage? Why would they prefer to sell at a lower price than a higher one? A higher forward price would incentivize reduced production, a lower one would not. Now they may restrict current production to increase the spot price and the forward price may be limiting how much they reduce production, but it would not be the lower forward price inducing reduced production.

manuelg writes:

Thought experiment: imagine you had a magical device, hooked up to a klaxon, that rang out before the market bubble burst. But, when it rang, you didn't know if the bubble would burst in 5 seconds, or in 3 months. It had some randomizer inside, and you didn't have access to know the setting.

There are some unknown number of such devices held by other investors. Maybe a dozen, maybe hundreds, maybe thousands. Each has a reason to keep the existence of their ownership of the device a secret.

You are waken at 3 A.M., by a terrible ringing. What do you do?

Every second you wait, after the klaxon rings, before the bubble actually bursts, you are making money. How long after the klaxon rings, do you pull the trigger?

The point is, even with this magical device, there would be a temptation to procrastinate on selling your whole position, there would be a temptation to try to time the market to the second.

Somebody, in possession of this wonderful device, will get caught with their pants down.

How the hell will this supposed science of bubble prognostication be any better than this magical device? It can only be worse, with worst results.

Bill Stepp writes:

But I do not think that bubbles are sufficiently detectable to make me want to have the Fed try to speculate against them.

The Fed causes bubbles trying to keep interest rates below their natural level. Would you have a fox guard a chicken coop? If not, then why should the Fed be allowed to speculate in a way that counters bubbles?

Matt writes:

Bubble precede change if we buy into the theory that bubbles are needed to cover the extra cost of reorganizing an economic sector.

If the software industry needs to reorganize development to cover the Internet, then we have two costs we are facing, one, the cost of getting some workers to leave their current job; and the cost of reorganizing those workers to fit another production system.

It is cheaper to do this all at once, creating uncommitted resources and reallocating them by mass auction, so to speak.

The bubble is needed to force the reallignment and pay for it.

Lord writes:

Producers could also buy futures and then reduce production to keep both up. Futures likely aren't that significant in themselves though since fluctuations in prices exceed futures spreads.

Krugman is correct though. On Wall Street, producers and consumers are not speculators by definition so their actions are not speculation.

Bob Knaus writes:

I'm confused. Isn't oil a consumable commodity, rather than a producing asset? Oil sitting in storage tanks doesn't create value, all it does is rack up storage fees. Aren't the assets the oilfields, refineries, pipelines, and other infrastructure required to produce fuel?

Seems to me that it's the same relationship as, say, corn and farmland. We witnessed an asset bubble in US farmland prices during the late 70s and early 80s, triggered by farmers' (and lenders') miscalculations around the permanance of the agricultural export boom.

Wasn't there a similar bubble in certain oilfield assets at the same time? The fact that we're seeing high oil and agricultural prices today, without accompanying bubbles in the assets that produce them, tells me there is no asset bubble. Farmers, oilmen, and their lenders all have long enough memories that they're determined not to get burnt a second time.

eric falkenstein writes:

I was a teaching assistant for Minsky in college. His basic idea, that people go from basic finance, to finance predicated on rising collateral prices, was kind of neat, but it really didn't have an empirical model. He just noted that many times, things overshoot. There was no testable model. It was, like Kindleberger's Manias book, basically descriptive.

I think it really has to be more specific about what causes a price increase to turn into a bubble, or a panic.

Mick Rolland writes:

The WSJ article is very thoughtful (or rather, the Princeton research): One of its most interesting points is that bubbles are hard to detect because overshooting price rises are almost accompanied by supporting fundamentals.

It is clear then, that judging whether prices in a sector are overvalued (a bubble) or undervalued is then a particularly difficult matter -and one not appropriate for the public sector. It should never be the goal of the Fed to alter specific prices in any particular market.

That said, if asset price bubbles are generated by both artificially low interest rates (mainly created by the Fed) and credit expansion 'out of thin air' (related to the former and to regulatory incentives), then the Fed's focus on 'core' CPI and short-term GDP disregarding inflationary effects in asset markets is too narrow.

The Fed should not prick "isolated" bubbles when they are not accompanied by credit expansion, but when there is a combination of accelerated credit growth and abnormally accelerated asset price rises, this should be taken into account and moderate aggressive expansionary monetary policies even if their short-term effect in the CPI is modest.

Asset bubbles have particularly deleterious effects on financial stability, and the Fed's present view that when a bubble 'pops' it should resort again to an expansionary stance (an assymetric reaction to bubbles) is highly questionable due to its inflationary bias and likely undermining of financial stability as it incentivizes overleveraging.

aaron writes:

I thought of another way prices can be inflated. The popular narrative the media uses is that "potential supply disruptions" drive up oil prices. Even if this doesn't describe the actual behavior of market participants, it may affect the end customer. It could essentially work as a marketing device, managing the consumers expectations and allowing prices to rise. It could be other way round; expected increases in gas prices are driving up oil as buyers compete for the higher margins.

However, I don't see how this can result in a bubble/pop. People aren't going to suddenly want less gas. If prices were to drop it would probably induce demand, without stock there could be runs and shortages. It's much more likely that people were previously paying much less than they were willing to.

But I do not think that bubbles are sufficiently detectable to make me want to have the Fed try to speculate against them.

What if instead the Fed simply designed markets so that they were less likely?

Economists and accountants live in a static world; there are very simple models that could be used to predict when changes in liqudity or money supply might lead to price instabilities. Surely something is better than nothing.

Comments for this entry have been closed
Return to top