A reader asked: “I came across an interesting article comparing the Nikkei 225 Index in the 1990s with the S&P 500 Index in the 2000s. Do you have any opinion on this study? Also I would love for you to post some data on how different asset classes performed in Japan from 1991 (the beginning of the ‘Lost Decade’) until now.”
The “Uncanny Relationship between the Nikkei and the S&P 500” identified in the first chart of the article echoes modeling done by Didier Sornette, Professor of Geophysics at UCLA, as summarized in “When Stock Market Models Crash”. His bubble decay model attempted to predict the behavior of the S&P 500 Index in the 2000s based on the behavior of the Japanese stocks in the 1990s. After two years of generally inaccurate live forecasts, he concluded:
“The origin of our failure to forecast the rebound of the market in 2003 can be traced back, in our opinion, to the following simple point: the mechanism underlying our projection is UNIQUELY based on the imitation and herding behavior of investors. It neglects the fundamental input of the economy as well as the external manipulations by the Fed and other institutions, the impact of foreign policy and foreign investors, the dollar effect (all these being possibly inter-related). It takes the extreme view that all these actions are endogenously determined and driven by the collective action of the investors. This is too simplistic because, in general, one can understand the evolution of [a] system as the result of an entangled combination of endogenous organization but also as a response to external news and exogenous shocks…”
He then revised his model of bubble decay to incorporate missing elements and developed a new forecast, which also does not do well (see the second chart in “When Stock Market Models Crash”).
The “uncanny” relationship portrayed in the first chart of the article appears to be founded on similar logic. Continuation of uncanniness is a possibility, but instead the past relationship may derive from substantial data snooping bias and thus break down out of sample. In other words, it may identify two patterns that match largely by accident out of a number of patterns investigated (and rejected) in undocumented searches. An out of sample test such as that conducted by Professor Sornette on his own model would be a reasonable next step.
There is a fair amount of financial markets research that supports belief in some degree of price and ratio reversion, as depicted in the second chart in the article. However, there is not enough information in the article regarding that chart to assess the statistical meaningfulness of the conclusion. How large is the sample? Does the sample appear to be normally distributed or wildly distributed? Is it reasonable to reach outside the sample period for the 1930 case without including other older cases?
It is more reasonable to view this article as posing hypotheses for testing than presenting a study.
CXOadvisory.com mostly focuses available time on U.S. markets as the largest, best documented and most scrutinized in the world. You can find much research on Japanese markets via searches within the Social Science Research Network (SSRN). For example, a search on “Japan lost decade” locates 29 papers.
Frequently Asked Questions
What is a bubble decay model in stock market forecasting?
A bubble decay model, as used by UCLA geophysicist Didier Sornette, tries to predict a stock index's path after a bubble by matching it to the behavior of an earlier bubble, such as Japan's Nikkei in the 1990s. It relies uniquely on investor imitation and herding behavior and treats market moves as endogenously driven by collective investor action.
Does Japan's 1990s Nikkei pattern predict how the S&P 500 behaved in the 2000s?
No, the Nikkei-based bubble decay model did not reliably predict the S&P 500 in the 2000s. Sornette's live forecasts were generally inaccurate over about two years, including a failure to anticipate the market rebound in 2003, and a revised model also performed poorly. The apparent match may reflect data snooping bias rather than a lasting relationship.
How does Sornette's model use Japanese stock market data to forecast the S&P 500?
Sornette's bubble decay model forecasts the S&P 500 Index in the 2000s by mapping it onto the behavior of Japanese stocks in the 1990s. It assumes the pattern is driven by investor imitation and herding. It was tested live, with forecasts made in real time and judged against subsequent market behavior over roughly two years.
Is the U.S. stock market in the 2000s really the same as Japan's stock market in the 1990s?
The U.S. market in the 2000s is not shown to be the same as Japan's in the 1990s; the apparent similarity between the Nikkei 225 and S&P 500 charts may be a coincidence. Continuation of the match is possible, but it may be two patterns matching largely by accident among many patterns searched and rejected. The comparison is better treated as a hypothesis for testing than as an established study.
What are the limitations of using herding-based bubble models to forecast stock markets?
Herding-based bubble models neglect fundamental economic inputs, Federal Reserve and other institutional actions, foreign policy, foreign investors, and the dollar effect. Sornette himself called the purely endogenous view too simplistic after his model missed the 2003 rebound. Comparisons built on past chart matches also risk data snooping bias and breaking down out of sample.