The Great China Bubble: Lessons from 800 Years of History
“When speculation has done its worst, two and two still make four.” -- Samuel Johnson
Summary: Clustering extreme returns and excessive turnover are hallmarks of a bubble. The average holding period of tradable shares in China is about a week, compared with a holding period of two weeks at the peak of Taiwanese bubble in 1990. Everyone is busy looking for the greater fool. As extreme returns with small possibility piles up, gains must accelerate to compensate excessive risk undertaking. But the probability to sustain such excess diminishes, too, much like being dealt straight flush every hand. History suggests that the coming six months will be a critical time window to monitor for potential crash.
------------------------------
A rational bubble: It is plain that China is in a bubble. With Shanghai rapidly rises above 5,000 points, a prevalent strategy amongst professional fund managers is to follow the crowd - but flee if the market ever slightly turns sour. Behold the world-record turnover that is more than double that of the US, and easily dwarfs any other markets. Exploding volume, concurrent with significant daily turnover, are the hallmarks of a bubble that is likely in its advanced stage. Now the question is how high it will go, how much longer before it go bust, and how will this fiasco end?
There are no easy answers. We are experiencing a rational bubble that arises from a rational valuation formula in modern finance. Recall that if we value a company’s stocks with perpetuity, the difference between discount rate and growth rate determines the company’s valuation multiple. Or simply P/E = 1/(r-g). The smaller the difference between the discount and growth rate, the higher will be the multiple. For instance, if the difference is 1%, than the company should be valued at 100 times.
In 1938, Von Neuman proposed an extreme scenario: what if discount rate equals growth rate? He posited that in a balanced growth economy, growth rate is always identical to interest rate, or discount rate. In this extreme scenario, stock’s value will always be infinite. And so comes the “growth stock paradox”. In an economy with rapidly falling interest rate such as China, it can be demonstrated that valuation multiple should be very high, or even infinite, with the gap between discount rate and growth rate narrowing quickly. As long as interest rate falls fast towards growth rate, valuation can be high even if growth continues to slow. This paradox seems to be able to explain the current dilemma: the rise in Chinese market has rational roots, but seemingly irrational outcomes.
Yet there are no magic beans in the market. We need to know where the boundary will be for market prices. Many before us had tried to define the boundary and failed. Just before the Great Crash in October 1929, Joseph Stagg Lawrence, a renowned Princeton professor, proclaimed that “the consensus of judgment of the millions whose valuation functions on that great market, the stock exchange, is that stocks are not at present overvalued”. “Where is that group of men with the all-embracing wisdom which will entitle them to veto the judgment of such intelligent multitude?” he inquired further.
While we all know how that movie ends, research in bubble has progressed greatly. Even so, we have not seen a model that can consistently predict the burst of a bubble – otherwise we wouldn’t be all sitting here. In science, we use “Occam’s Razor”, or the Law of Parsimony, to determine the practicality of a theory. The simpler the theory, and the less variables to control for future prediction - the better. For this purpose, the famous LPPL model is a little too complicated, with somewhat sporadic prediction success. Rather, we will look into history to find some clues to the future.
Clustering extreme returns, excessive turnover are hallmarks of a bubble; the coming six months critical: We have investigated a combined 800 years of global financial data, from 17th century UK stock market, to 18th century gold and silver prices, and onto 19th century US stock market to discern the ordinates within bubbles. Our quantitative research shows that a typical bubble tends to take around five years to develop, and exhibits clustering extreme returns when approaching its peaks. In
47 31211 47 14918 0 0 1496 0 0:00:20 0:00:09 0:00:11 3034 general, when return consistently exceeds two standard deviations from its long term logarithmic return, a bubble is forming. And it normally takes about another six months before the bubble deflates – or even burst. With this methodology, we have identified 26 bubbles, 21 incidences are correct.
The following charts show the relationship between extreme returns and China’s market index (Focus Chart 1). We can see the observations aforementioned clearly demonstrated in these charts: 1) clustering extreme returns leading the way to the peak of bubble; 2) the occurrences of extreme returns tend to happen approximately 6 months ahead of the peak; and 3) these extreme returns belongs to the top 5% of the return distribution.
Focus Chart 1: Extreme returns tend to cluster 6 months before peaks; 95% of return is lower than currently observed.
The reason for clustering extreme returns nearing the peak of a bubble is intuitive. Traders are constantly assessing whether to stay in the market by comparing the return of holding stocks into the next period versus cashing in now. As the market rises, higher returns become less likely on the distribution tree. Thus, the expected return that traders require to stay in the market must soar to compensate for the ever dwindling chance of winning it, as seen in the near-vertical climb of the market index. However, as these small-probability events continue to accumulate near the peak, the probability of further gains grows ever smaller – much like being dealt a straight flush every hand - till the probability becomes so small that a crash arrives inevitably.
If this explanation holds water, then we must see market turnover accelerating concurrent with extreme return registration nearing market peaks - as it is now. This is because traders start to constantly assess the risks with their positions within an ever-shortened time span – attempting to time their exit with perfection. If accelerating turnover and clustering extreme returns are concurrent, it should lead the peak by around six months as well. This is indeed the case (Focus Chart 2).
When calculated on a free-float adjusted basis, Chinese market’s average holding period is about one week – a hallmark of intense speculative trades in the market. Everyone is busy looking for the greater fool. Note that at the height of the Taiwanese bubble in 1989, every available share on the exchange changed hands close to twenty times per annum. That is, the free-float shares on Taiwanese exchange changed hands every 15 days on average.
Focus Chart 2: Free-float-adjusted shares turnover almost 3 times/month; average holding period is about 1 week.
Shanghai can climb to 6,100: Where would the Chinese market peak? This is a more difficult question to answer. A Chinese market adage dictates that one shall never predict the peak index level during a bubble. Sir Alan Greenspan once said, “Bubbles generally are perceptible only after the fact. To spot a bubble in advance requires a judgment that hundreds of thousands of informed investors have it all wrong. Betting against markets is usually precarious at best.” Most would rather let their reputation “fail conventionally than succeed unconventionally”.
We have had our fair share of failure in predicting the peak of the market this year. When we wrote our outlook piece titled “China 2015: A Bubbly Nation” for Bloomberg on January 29, we foresaw a bubble forming in the Chinese stock market in 2015. Yet we had set our index target too conservatively at 4,200 by comparing bond yield versus earnings yield, even though the market was at around 3,100 then. After the market blew past our target of 4,200, we recognized the strong upward thrust, without setting a specific target.
This time around, instead of using the traditional PE, PB multiples to value the market, we have used the Market Cap/GDP ratio. We note that the market was willing to pay ~1.2 unit for each unit of GDP earned at the peak of the 2007 bubble. This ratio, observed at the peak of the bubble, is a reflection of market sentiment, and can be used as a reference to evaluate the current bubble (Focus Chart 3). Furthermore, in other countries such as the US, this ratio is mean reverting. If this relationship continues to hold in the current market cycle, then the Shanghai Composite can rise above 6,100.
Focus Chart 3: Market Cap/GDP is soaring, and is approaching its peak in Nov 2007. This ratio is mean-reverting in the US.
Monetary policy will continue to ease: Incidentally, China real effective interest rate, after taking into account consumer and house price inflation, remains elevated despite rounds of interest cuts. That is, China will continue down its path of monetary easing to really lower the interest burden of the Chinese economy. The real effective interest rate tends to lead the market by around six months as well, coinciding with the time frame identified in the variables discussed above (Focus Chart 4).
Focus Chart 4: Real rate still elevated, hinting at further easing; It leads the Shanghai Composite by ~6 months.
The Great Crash in 1929 has demonstrated bubble bursting is notoriously difficult to time, and a financial collapse has seldom happened in bad times. For instance, macroeconomic flows were in good shape just prior to the crash. Hoover’s election into the Office prior to the crash was also supportive to an optimistic political mood. Indeed, the preeminent Harvard Economic Society proclaimed a few days after the crash that “a severe depression such as 1920-21 is outside the range of probability. We are not facing a protracted liquidation.” Such disastrous preordination eventually led to the closure of the society later in 1932.
On Black Monday in 1987, the Dow fell 200 points soon after open but then recovered back to above 2,100. It appeared as if the Dow could escape with a loss of only about 200 points, till at 2:45pm a new wave of massive sell-off started, sending the Dow into a free fall of 400 points loss at closing. Yet that is not the end of the saga – later on it was discovered that because of the extremely heavy volume, the exchange’s computers were running hours behind trading. And the final loss tally exceeded 500 points, or 22.6% - the largest single day percentage fall ever.
Clearly, bubble crashes have eluded many in the past, and will continue to do so in the future. In this short paper, we have perhaps barely scratched the surface. There is no easy answer to the situation that we are facing within such limited space. But we should take Keynes’s advice to “leave to Adman Smith alone of the quarto … pluck the day, fling pamphlets into the wind, write always sub specie temporis.” Or as Dylan sings: the answer, my friend, is blowin’ in the wind.
Hao HONG, CFA
2015-06-16 03:00am