But Earns Partial Credit from Blogging Physicist!

We've been following the story of Didier Sornette and his colleagues' prediction that the Shanghai Stock Exchange was in the throes of a building bubble, which would soon implode, most likely between the dates of 17 July 2009 and 27 July 2009.

Well, those dates came and went, and as noted by the intrepid blogging physicist of MIT's Technology Review arXiv blog, KentuckyFC, the hypothetical bubble didn't pop.

In July, he and his buddies pointed out that the Shanghai Composite stock market index was following exactly this kind of trend. But they also made an extraordinary prediction. They said that this bubble would burst between July 17 and 27.

That's a very specific prediction of the kind that economists almost never make. How they came to their conclusion wasn't clear, and I, for one, was very skeptical. In fact, I bet he was wrong and promised him an arXivblog T-shirt and baseball cap if the market proved otherwise.

So I kept an eye on the index, and on July 27 noted that it was still going strong. In fact, between July 17 and 27, the index rose by 251 points, or about 8 percent. So much for the crash.

So you might think. As we noted in originally commenting on Sornette's prediction however, those were merely the dates he and his colleagues indicated would be the most likely for which stock prices might begin falling. The full range of dates for which they indicated such a fall might occur ran from 10 July 2009 through 10 August 2009.

There was no crash between 17 July 2009 and 27 July 2009. But what about the days from 27 July 2009 through 10 August 2009:

Then something strange happened. On August 4, the market hit a peak of 3,471, and then it dropped. Dramatically. By August 19. it had fallen to 2,786, a drop of about 20 percent.

Several questions come to mind. Is this the fall that Sornette and company were predicting or just a coincidence, a regression to the mean? And if the fall is related to their prediction, could the drop have been caused by it?

There's no way of knowing, really. But it's too close to Sornette's original prediction to be ignored. If he has found a way to predict (or trigger) crashes of one kind or another, then it's hard to underestimate the significance of such a breakthrough.

We can answer some of these questions. First, no, Sornette and his team did not trigger the 20% fall in the Shanghai Stock Exchange (SSE) index with their prediction. In our previous coverage of this story, we noted that Sornette had erred in identifying the relatively rapid rise in stock prices on the SSE from November 2008 into July 2009 as evidence that a bubble had formed in that stock market.

Instead, we had found that the growth of the SSE's stock prices were primarily driven by the expectation that the business situation of Chinese companies was becoming less bad. Given that we've seen the same, mostly parallel dynamic in the U.S. stock markets since they bottomed on 9 March 2009, this close coupling of change in stock prices with the observed change in fundamental business performance provides a clear indication that a bubble had not formed on the Shanghai exchange, since a bubble may be said to exist in a stock market if, and only if, the value of stock prices have become decoupled from the value of their underlying dividends per share.

Recognizing that reality, we offered a prediction of our own:

That's not to say that stock prices might not crash on the exchange. Instead, if they do, it will be because of fundamental changes in the outlook for the companies on the exchange, which would be driven by factors that would change the amount of dividends they pay to their shareholders in the future. Not a fun situation, but still, not the result of having a bubble form, then pop.

As it happens, that's *exactly* what happened to the Shanghai Stock Exchange index following it's peak on 4 August 2009 (or more accurately, beginning on 6 August 2009). We found the fall in the value of the SSE index coincided with a very definite change in the amount that a Canadian-based insurer with a very large exposure to China's economic situation would pay out in dividends per share to its stockholders going forward, Manulife.

We see this outcome directly in our chart showing the relative change in the SSE index, the S&P 500 (for reference) and the value of Manulife stock (MFC). We see the result of Manulife's announced halving of its dividend payment on 6 August 2009, which directly coincides with the opening gap we observe between the S&P 500 and the SSE index at that date. Moreover, we see that MFC's change in stock price has largely preceded and marked the boundary for the change in value that we observe in the SSE index.

Finally, we observe that both MFC's and the SSE index finally bottom and begin largely moving in tandem again beginning 17 August 2009 - we do note the SSE deviated from that track on 19 August 2009, which appears to us to be the result of unique, short-lived noise within China's stock markets.

Being the result of a defined change in investor outlook, we can completely rule out the Sornette team's prediction as being a trigger for the event. Looking at KentuckyFC's other questions, we can also largely rule out the possibility that the change in value of the SSE index is the result of reversion or regression to the mean - what we observe instead is consistent with a "unit root process," where a shock (in this case, a change in fundamental investor expectations) has a permanent effect that effectively "resets" the level of stock prices. This differs from what we would call a "noise-driven event," such as we observe in the SSE index on 19 August 2009, where stock prices will resume pacing a given trajectory.

The only question we can't answer is the one that asks whether the crash in stock prices that we observe in the SSE index in August 2009 is the one that Sornette's team predicted or just a coincidence.

That will require a new batch of predictions.

In the meantime though, the prediction Sornette and his colleagues did make is enough to claim partial victory - and a free arXivblog hat from our blogging physicist!

Labels: forecasting

- posted by Ironman at 12:49 PM | Permalink | Home |