« Did the structure of banker pay cause the crisis? | Main | Assorted links »
What it means to predict a crisis
Some economists are trying to get macroeconomics off the hook by arguing that by their very nature crises are unpredictable. Thus David Levine aggressively argues that "our models don't just fail to predict the timing of financial crises - they say that we cannot."
There are three problems with this argument. First, it assumes what is it at question - namely whether what Levine calls "our models" are good models. Perhaps behavioral models could better predict the timing of financial crises. I will not push this argument but I do believe that current events call for a greater than normal willingness to think beyond the confines of the models that one defends.
Second, it's not true that "our models" tell us that we can never predict a financial crisis. In some cases, our models predict the exact moment that a crisis will occur and these models are perfectly consistent with, indeed require, rational expectations. It is perhaps no accident that Paul Krugman has specialized in these types of models.
Third, the word timing is misleading. Let's accept that a crisis cannot be predicted to the day or even to the year. Nevertheless, it is perfectly reasonably and fully consistent with rational expectations to predict an increased probability of a crisis.
If you play Russian Roulette with 1 bullet and 100 chambers in your pistol, I can't predict when the crisis will occur. If you play with 10 bullets, I still can't predict when the crisis will occur but I can say with certainty that the risk has increased by a factor of ten. Analogously, nothing in modern economics makes it theoretically impossible to forecast that greater leverage and higher than normal price to rental rates, to name just two possibilities, increase the probability of crisis. Nor does modern theory make it theoretically impossible to forecast that conditions are such that if a crisis does occur it will be a big one.
All of this is true even in the context of stock markets. Efficient markets theory implies that any two stocks will have similar risk-adjusted returns it does not imply that the risk of bankruptcy is the same for any two firms. It is perfectly reasonable to say that Google revenues are going to have to increase at a historically unprecedented rate or the stock will plummet. It is even consistent with efficient markets theory to predict that the probability of Google stock falling is much greater than the probability of it rising (but if it rises it will rise very far, very fast).
Thus the "we could not have predicted the crisis even in theory" argument is a weak defense--even with rational-actor, rational-expectations models there are plenty of senses in which economists could have better predicted the crisis and, although this is yet to be seen, perhaps they could and will do even better with other sorts of models.
Posted by Alex Tabarrok on September 21, 2009 at 07:44 AM in Economics | Permalink
Comments
> It is even consistent with efficient markets theory to predict that the probability of Google stock falling is much greater than the probability of it rising (but if it rises it will rise very far, very fast).
easy to test whether the market believes this: look at the options markets
Posted by: babar at Sep 21, 2009 8:26:48 AM
Great comment.
Another point might be worth considering as well. Water has three states: liquid, solid and gas, and how it behaves in each state is different than the other state.
Economists seem to assume that the economy is at equilibrium or that equilibrium is close at hand or that the economy, when it is wildly out of equlibrium, obeys the same forces as when it is in or close to equilibrium.
Economists have been unwilling to concede to another camp that is certain disequlibria moments that other rules may apply, or rules that have little force and effect normally have much greater effect in another state.
Nor do economists model irrationality very well, particularly when it conflicts with their theory. But, we know people are predicatably irrational, and that is why there is such a thing called marketing. So, why can't we accept that there might be different behaviours of human beings--in response to declines in savings savings, shocks to wealth, job security--that prompt unexpected behaviour? If we model more broadly, we might find that there are many possible alternative states that are possible from one policy, and that different theories apply better under different conditions.
That is not to say that we can't exclude some policies as being unwise, even though there may be many just as successful alternative policies. Let me give you an example: there is some criticism of game theory that it is not practical because there are multiple possible equilibria. As a lawyer defending someone accused of price fixing, that is great for me: I can use game theory to show that there are altnernative non-collusive reasons for a given equilibria: I am using game theory to show something is just as probable, or that something is more probable. Economics can be used to exclude arguments, or include more arguments or alternatives.
The same is true in macro. You might be able to exclude somethings as policies you wouldn't do during a recession, while leaving people to debate alternatives based, really, on their tastes and preferences and not so much on the high probability that one theory is the only theory that will work. The world is complex.
Posted by: Bill at Sep 21, 2009 8:30:50 AM
Your number 3 is the biggie. We can't predict when excessive speed or drunken driving will cause a crash, but we do know that each makes crashes more probable. Predicting the timing of financial crises is beside the point, what we want to do is limit behavior that promotes or aggravates such crises.
Posted by: capitalistimperialistpig at Sep 21, 2009 8:36:36 AM
>If you play Russian Roulette with 1 bullet and 100 chambers in your pistol, I can't predict when the crisis will occur. If you play with 10 bullets, I still can't predict when the crisis will occur but I can say with certainty that the risk has increased by a factor of ten.
I dont know jack about economics, but this argument is dangerous and misleading. Because real world is not like a casino where you know what all the outcomes are as in an artificially constructed game. (Ref: Taleb, The Black Swan).
Maybe at best you can work on small scales like Google's stock prices maybe. But if similar arguments are used in a much broader and more complicated system, it will only give a false sense of security. The limitations of the models have to be kept in sight always. My two cents.
Posted by: milieu at Sep 21, 2009 8:40:52 AM
On a different but related note, the models used to derive pricing for securitized mortgages made a spectacularly wrong prediction: that housing markets would remain local in nature and a downturn in one market would likely be offset by an upturn in another market.
So far, I haven't seen any good explanation of how and why they got this wrong. (Bad explanations are along the line, we didn't have enough data.)
Posted by: michael webster at Sep 21, 2009 8:56:45 AM
Most crises, whether financial or political, are preceded by a series of warnings. The fact that most of us did not pay enough attention to the warning signs does not mean they weren't there. They were. Plenty of them.
Posted by: Pavel Kohout at Sep 21, 2009 9:07:39 AM
Do you or Tyler have any commentary on this:
http://leiterreports.typepad.com/blog/2009/09/alex-rosenberg-on-cochrane-and-economics.html ?
Posted by: JW at Sep 21, 2009 9:11:13 AM
Pardon me, but I realize I was responding to a different post--the one about conflicting schools of economics--rather than this post. What I said has some relevance to prediction, but I apologize for being of subject.
I agree with Pavel. Financial crises are preceded by warnings. But, politicians don't respond to warnings, they only repond to crisis because that is the only way to get everyone focused.
Posted by: Bill at Sep 21, 2009 9:13:10 AM
Very simple: economy it's not an exact science, as mathematics, that is, with a primary formula you could preview the future and solve any economics fact.
Posted by: audiokabel at Sep 21, 2009 9:53:37 AM
It does seem that markets are more prone to tipping points then most models predict. I thought the tech market was over priced and didn't understand why so many were still pushing the market higher. The risk premium had turned negative, But for reasons that I don't fully understand, rather then the market adjusting slowly, it built to a tipping point and then fell over.
In the current financial crisis I think most fail to appreciate the impact of the oil price shock which has the kindling for the current crisis. People who were stretched to make loan payments were hit with a large jump in fuel prices. A weak auto industry took another severe blow from increased oil prices. At the same time Obama took a commanding lead in Presidential polls, while talking about increased regulations and higher taxes.
During this time the Fed was talking about raising rates to fight inflation. Why shouldn't bankers be confused about the future. Yet they continued to be overly optimistic.
Still, over the last few years, I was shocked at housing prices in Miami, Arizona, and Nevada. I was shocked at the increases in housing costs in a normally sane markets like Chicago. Expectations of future housing prices started to reach Ponzi levels. Low interests rate encouraged people to over invest in housing. Normal prudent guidelines were ignored. But was it irrational? No. Assumptions about the future were clearly wrong but many people did very well. People can be rational and wrong
Posted by: DanC at Sep 21, 2009 10:15:32 AM
HFS, I agree with capitalistimperialistpig.
Posted by: Andrew at Sep 21, 2009 10:30:03 AM
Great post Alex. I've said elsewhere that this notion--"the EMH is fine because the EMH says you can't predict crises"--is like fundamentalist Christians saying, "The Bible is trustworthy because Paul writes in his letter to the Corinthians..."
Posted by: Bob Murphy at Sep 21, 2009 11:14:35 AM
Predictability (or its lack) is certainly the economic 'flavor of the month'. It's all over the blogs and even Krugman has gotten into the act.
It's funny watching the macros trying to absolve themselves, when anyone who follows yields (and yield curves) could have predicted this crisis years ago.
For instance, Fannie Mae and Freddie Mac were of questionable solvency when house prices were on a tear in 2002 and 2003 and money costs were near 1%. Figuring the spreads on the back of a piece of scrap drywall, it would be clear that both - the GSE's and the real estate industry that depended utterly upon them - would be kaput if the Fed Funds rate exceeded 4 percent.
It did and they did. Very predictable. In 2005, I was telling people that the real estate boom was over and for the highly leveraged to get out. Nobody listened to me and a lot of people I know are busted. This is why my brother introduces me as, "my brother who predicted the banking crisis five years before is started".
George the Electrician predicted the financial crisis long before, as well. "It's the American way," he said, "to drive an idea into the ground then walk away from the wreckage and go somewhere else and start over."
Neither Paul Krugman nor the rest of his macro compatriots got it. This crisis was a no- brainer. What's also a no- brainer is that the current crisis will get worse. In the background are all the same 'missed the boat' economists calling for a (speedy) recovery. Wrong again. Yields are the indicator again as the banking system is flickering near death even when money costs are near zero. What happens when yields start to rise - an indirect result of the flood of liquidity pouring into the banking and finance sector?
Answer is a return to bankruptcy with a vengeance. If the system is degrading while cheap debt is raining down like manna, what does the establishment have as an alternative when the off- balance sheet bad loans start to emerge onto the market looking for a buyer, any buyer? The Fed and the Treasury have nothing new other than what they are already offering; cheap dollar loans.
The finance/loan bubble represents the percentage of bad loans the government can zombify. It can't zombify all of them; the government has one foot in the real economy and the zombie process reduces the value of the dollar, which is also used to purchase the 10 million barrels of crude oil that are imported into America every single day.
The price and availablity of crude oil is the connecting factor whereby finance aims to swamp the productive economy. Not enough crude or too high a price in dollars and there is no economy, period! Right now, the price of oil is too high to support ANY productive growth; almost all that is being currently registered worldwide is central- bank- bloated price inflation of speculative assets such as stocks. This contrived 'growth- lite' is a failing hedge against the high price of the resource necessary for top line commercial solvency. The earnings are only for speculators, not for producers.
It's as if a region hopes to profit by a gambling casino - where some neighbors take money from their neighbors - while all the factories and shops in the region close their doors. This is the 'real paradox of thrift': eventually, the lucky - or cheating - gamblers take their winnings home - to the Turks and Caicos. The rest are left destitute.
The productive economy lacks the physical means to service then retire all the debt that overhangs it. The outcome is deflation, but the mainstream modelers don't recognize this. They also miss the energy problem, the debt overhang problem, the trapping of liquidity within finance problem, the inability to earn/service and retire debt problem ... and will with certainty miss the next deleveraging leg.
It will comes as a great surprise to them and the 'Black Swan' will fly again. Good grief!
Posted by: steve from virginia at Sep 21, 2009 11:38:28 AM
Alex, interesting post. to address your second and third points:
2) I admit that I'm not intimately familiar with all of Krugman's work, but from the summary that you linked it looks like his model is entirely deterministic. In other words, if everything in the future is known with certainty, we can predict when a crisis is going to occur. To be fair to Paul, I don't think that was the point of his paper, but that's what you seem to be taking away from it. Once you start making the model more realistic (and more interesting), you lose this feature. This leads to the third point:
3) I think you may have missed this line from Levine's essay: "Just as those models predict only the statistical distribution of photons, so our models only predict the likelihood of downturns - they do not predict when any particular downturn will occur." How is this any different from what you are suggesting?
Posted by: izzy at Sep 21, 2009 11:57:05 AM
Yes the crisis was entirely predictable, except for the timing. So are the dozen other predictions of how the dozen other nesting bubbles are going to pop. The Russian Roulette example implies a bearish viewpoint which is that all predictable bad events are going to happen if we keep playing. It may even be true, but who was it who said that the market can stay irrational longer than you can stay solvent betting against it? The timing and time scale is very important, and it doesn't even take an economist to see many of the looming potential disasters.
But in the meantime, there's bonuses to be made and elections to be won.
Posted by: Josh at Sep 21, 2009 12:06:39 PM
Thus the "we could not have predicted the crisis even in theory" argument is a weak defense--even with rational-actor, rational-expectations models there are plenty of senses in which economists could have better predicted the crisis
If you believed in EMH, then predicting the crisis was trivial. People claimed to have found a way to create ever-increasing returns with no risk. This can't happen, according to the EMH. Therefore, they were hiding the risk somewhere. VaR said that they weren't hiding the risk in the likely events; therefore, they were hiding it in the tail events.
I dont know jack about economics, but this argument is dangerous and misleading. Because real world is not like a casino where you know what all the outcomes are as in an artificially constructed game.
A casino does, however, make for an excellent analogy to what was going on. Consider the martingale betting strategy. This is the simply strategy in a game where whenever you lose, you double your bet until you win. Then you start over with a lower bet. It sounds foolproof (and the mathematics required to show that it is not are fairly deep), but it inevitably leads to ruin. While it is true that eventually you'll win after every run of losses, it's also true that eventually you'll have a streak of losses so big that it exhausts your bankroll/line of credit. What happened to many in the crisis was simply an example of this. (It's also largely what happened to LTCM.)
Posted by: John Thacker at Sep 21, 2009 12:08:28 PM
So macro cannot dependably warn society of an impending crisis. What can it do?
Posted by: Thomas DeMeo at Sep 21, 2009 12:50:11 PM
I didn't read all of the comments. But it seems kind of like you didn't read the whole article, Alex. Levine writes that it does have predictive power, just that it won't 100% of the time be accurate.
Posted by: Paul at Sep 21, 2009 1:06:37 PM
I am glad that Steve from Virginia, clearly a multimillionaire because of his ability to predict obvious market errors, has taken time away from his world tour to tell us how we can become millionaires like him.
Or did you lack the courage of your convictions to actually bet your retirement money on what was so very obvious?
Posted by: DanC at Sep 21, 2009 1:55:47 PM
I approach the problem from a different direction: If we could predict a financial crisis with 100% accuracy, repeatedly, the predictive tool would have to be useless, because any action made based on the prediction will also change the market, therefore altering the market. The market's behavior whenever 'it' knows that the crisis is looming would do nothing but accelerate such a crisis, as the markets adjust to the new perceived value of the securities involved.
So at one point, even a theoretically perfect model stops predicting, and eventually starts triggering the crisis, making itself useless.
Posted by: hibikir at Sep 21, 2009 1:57:32 PM
Also Steve from Virginia, read the John Thacker post. If you saw the problem five years ago, and had bet against the market five years ago, unless you have very deep pockets, you would have gone broke before your bet would have paid off. Some others went broke thinking the market should have turned sooner.
Saying that the crow can only fly so high doesn't tell you when he will peak and descend.
Posted by: DanC at Sep 21, 2009 2:03:53 PM
To follow Mr Demeo, economists seem to be telling us that economics is useless in the real world. I knew that.
Posted by: Robert Speirs at Sep 21, 2009 2:33:45 PM
This whole "The market can stay irrational longer than you can stay solvent" line is pretty much BS. That is only true if you use leverage. There are lots of non leveraged ways to bet against the market.
First and easiest is get out of stocks if you are in them. You can hold that position forever.
Second if you want to go further short sell S&P 500 futures against the cash you have in the bank making sure that you allow for a considerable cushion for margin calls. For instance make sure that you have enough cash to cover margin calls over ten years equal to the greatest 10 years ever times 2. Unless the market grows at twice the amount it has ever grown over the next ten years you can hold your position for at least 10-years before your capital is exhausted.
You can also buy one year ATM puts with 10%
of your capital each year for ten years.
Both of these strategies allow you to ride out any irrationality for at least 10 years while betting on a drop.
So if folks "absolutely knew we were in a bubble" and it was "clearly evident" their are plenty of strategies to bet that way with a ten year time horizon.
Also you could have sold your house and rented locking in a long term lease.
Posted by: eccdogg at Sep 21, 2009 2:46:55 PM
Or simply hedged. Or just not bought a house in Vegas. If people get the idea of limiting bets it's not apparent from the rhetoric.
"If you didn't go 30 to 1 leverage on S&P ultrashorts, you don't really believe what you are saying." Well, why not 40 to 1?
There isn't much shame in missing it or not going all-in. Phil Fisher is one of the best market timers on the planet. A pro. He didn't see it coming. Others did. If you bought stocks 10 years ago, you are about even. If you bought them thinking the duration of stocks is a couple years, that's a portfolio policy problem.
Posted by: Andrew at Sep 21, 2009 3:25:57 PM
I meant Ken Fisher, the son.
Posted by: Andrew at Sep 21, 2009 3:26:59 PM
