Unconventional monetary policies of the Fed et al. are having distortionary effects on asset markets and increasing their inherent instability. Via a mathematical framework, that has been empirically shown to accurately model bubbles, and give a prediction for when they will ultimately come to an end, Variant Perception (VP) investigates the S&P 500 today. The framework suggests a window of time within which the S&P will experience a ‘regime change’, which may involve a steep price drawdown. Using this framework, VP predicts this regime change will occur any time between now and the beginning of Q413. Furthermore, based on a suite of indicators for the S&P which point towards overbought and overvalued conditions with a backdrop of complacency and divergent market internals, VP forecast any regime change to resolve itself through a – potentially steep – market drawdown. Remaining long is an ugly wager.
> The US equity market is likely close to the end of a bubble phase that started in 2009. The bubble pattern may resolve itself by trading sideways, but it is more probable we are facing a significant drawdown.
> The critical window for the end of the bubble is anywhere between now and the beginning of 4Q13. Determining when a bubble might burst is very difficult. The end date for bubbles is sensitive to initial conditions of the fitted model. However, looking across the models we have fitted we are now in the window of the critical period of the current S&P bubble.
> Buying put spreads or protection makes sense given the asymmetric outcomes. The chance for further upside is small versus the larger probability of a market drawdown that could be 10% upwards. Volatility is cheap and under-priced. Investors have rarely bought so many calls versus puts, and volatility is mispriced.
> The current belief is that central banks can backstop risk taking. Bubbles are often formed by a new belief or paradigm. Today everyone thinks they can own stocks to benefit from “QE Infinity” while the Fed provides liquidity. Many investors believe they will be smart enough to get out and reduce exposure before the Fed changes its stance. We doubt everyone will be able to make it through the exit doors at the same time.
> High margin debt is in line with 2000 and 2007 peaks in absolute terms. In year over year terms, the growth in margin debt is near 1987, 2000, 2007, 2010 and 2011. In most of these cases, these indicated intermediate market tops and in many cases preceded crashes. Previous increases in margin debt in the past three years contributed to the Flash Crash of 2010 and the debt ceiling crash of 2011. High levels of margin debt signal that a disorderly resolution of a bubble is more likely.
> Our short-term sell signals are almost all flashing sell, and we have a breakdown in market breadth and health. Our collection of buy and sell signals have flashed many sell signals and no buy signals. Investors are very long and complacent, Treasury yields and oil prices are rising, and we are seeing deterioration in market breadth.
> Valuations before drawdowns tend to be high, but they don’t need to be ridiculously high. Current valuations are high and problematic. US markets appear cheap on the basis of forward-operating earnings. However, almost all US indices are near the top end of historical valuations based on trailing 12 month earnings. Trailing earnings are more accurate indicators of future returns
BELIEF IN CENTRAL BANKS CAUSING BUBBLES AND CRASHES
Every year, the Darwin Awards are given out to honour fools who kill themselves accidentally and remove themselves from the human gene pool. The 2009 Award went to two bank robbers. The robbers figured they would use dynamite to get into a bank. They packed large quantities of dynamite by the ATM machine at a bank in Dinant, Belgium. Unhappy with merely putting dynamite in the ATM, they pumped lots of gas through the letterbox to make the explosion bigger. And then they detonated the explosives. Unfortunately for them, they were standing right next to the bank. The entire bank was blown to pieces. When police arrived, they found one robber with severe injuries. They took him to the hospital, but he died quickly. After they searched through the rubble, they found his accomplice. It reminds you of the immortal line from the film The Italian Job where robbers, led by Michael Caine, after totally demolishing a van in a spectacular explosion, shouts at them, “You’re only supposed to blow the bloody doors off!”
Central banks are trying to make stock prices and house prices go up, but they will likely ‘blow the doors off’. Unconventional monetary policy tools are intended to generate spill-overs to other financial markets. For example, QE and central bank asset purchases are meant to boost stock prices and weaken the dollar, lower bonds yields and push investors into higher risk assets. Central bankers hope they can find the right amount of dynamite to blow open the bank doors, but it is highly unlikely that they’ll be able to find just the right amount of money printing, interest rate manipulation and currency debasement to not damage anything but the doors.
Bubbles are often formed by a new belief or paradigm. Today everyone thinks they can own stocks to benefit from “QE Infinity” while the Fed provides liquidity. Faith in central banks today is equivalent to faith in the world of dot-com in 1999 or faith in the eternal rise of housing prices in 2006.
Targeting stock prices is par for the course in a world of unconventional monetary policy. Officially the Fed receives its marching orders from Congress and has a dual mandate: stable prices and high employment. But in the past few years, the Fed has unilaterally added a third mandate: higher stock prices. Chairman Bernanke himself pointed out that stock markets had risen strongly since he signalled the Fed would likely do more QE during a speech in Jackson Hole, Wyoming in 2010. “I do think that our policies have contributed to a stronger stock market, just as they did in March of 2009, when we did the last iteration [of QE]. The S&P 500 is up about 20% plus and the Russell 2000 is up 30% plus.” It is not hard to see why stock markets rally when investors believe the most powerful central banker in the world wants stock markets go up.
The problem is that almost all previous ends to QE or fears of an end have resulted in steep drawdowns. Almost all clients we speak to believe they will be smart enough to get out and reduce exposure before the Fed changes its stance. We doubt everyone will be able to make it through the exit doors at the same time. The last two times the Fed stopped QE or paused, we had significant price declines: Flash Crash May 2010 and August 2011 with the debt ceiling. For example, the Flash Crash in 2010 happened at the end of QE1. The August 2011 drawdown happened right at the end of QE2. It is highly likely we’ll see a repeat when the Fed tapers.
AN ANALYTICAL FRAMEWORK FOR BUBBLES
Here we introduce a framework for modelling bubbles. We have left out the mathematics. Instead, without getting too bogged down in the detail, we hope to show that it is possible to model bubbles using a theoretical approach, and that this approach can yield an estimate for when the bubble will end. We then show that the S&P today is exhibiting bubble-like behaviour and, using the model, give an estimate, of between around now and the beginning of 4Q13, for when the bubble will come to an end.
Sornette’s central finding is that that log-periodic power laws (LPPL) do a good job of describing speculative bubbles. An LPPL essentially attempts to model the price of an asset as a function of oscillations around an exponential growth rate.
IS THE S&P 500 IN A BUBBLE?
In the section below, we try to answer this question applying the method described above with a heavy emphasis on exploring what information the LPPL model can currently give us on the trajectory of the S&P 500.
The ensemble of critical times from our analysis suggests that the S&P 500 may be exhibiting characteristics of two bubbles and both are at or close to critical points. Critical times for windows starting prior to February 2011 range from 1 to 4 days. This would correspond to a longer duration bubble with origins in 2009 that has reached a critical point. Critical times for windows starting from 2012 onwards show a wider spread of critical times, ranging from 24 to 100 days (with a mean of 63). This would put a critical date on a more recent and shorter duration bubble at some point in the beginning of 4Q13.
Full Variant Perception research note below: