The rise of algorithm-driven trades carries risk, especially when a price correction may be afoot
Bubbles, I've seen a few, which swelled and popped in familiar cycles during my 20-odd years in the finance industry.
One arose from a belief that the Asian miracle was unassailable despite ridiculous valuations and a broad gap between savings and investments; another was due to a dot.com frenzy; and the most recent was a global housing-market bubble. The mantra in the early 2000s was "buy land, they don't make it anymore" - which led to an all-mighty pop that triggered the 2008 global financial crisis.
The global economy grew at 3.1 per cent last year, and the International Monetary Fund (IMF) expects the rate to accelerate modestly to 3.4 per cent this year. This would still fall well short of the pre-crisis average of 4.3 per cent.
Yet, today, most global stock markets are again pushing historical highs, even as the world economy struggles to hit escape velocity and conclusively consign the global financial crisis to the history books. The more things change, the more they remain the same.
Nine years after a bubble bursts, a new one invariably starts to emerge.
My small talk at home with neighbours, on the plane with stewardesses, in gyms with health freaks and at restaurants with waiters is no longer about which property or stock to buy, but about bitcoins, makers of driverless cars, the remarkable business models of Amazon and Alibaba, the resilience of the market since Brexit and United States President Donald Trump's surprise election win, and the low administrative costs of exchange-traded funds.
All see a reason to put money to work at every turn. It's different from pre-2008 days, but alike too - every dip is a chance to buy.
What's new is the prevalence of an algorithm- or computer-driven trading style that has emerged in recent years, to the dismay of macroeconomic-based market strategists like myself. The new style, engineered by computer whiz kids, is the flavour of the month.
Estimates of how much money had been parked with such algorithm-based funds by investors as at the end of last year go as high as US$2 trillion (S$2.8 trillion). These days, market influence seems to have swung from one category of nerds to another.
In the region, Singapore is at the forefront of the move towards trading activities driven by financial technology, or fintech. But the impact of such a sea change has yet to be fully understood globally. The risk of a major market surprise is high, in my view. Fed by a post-crisis trend of shallow price dips coupled with ample global investment liquidity, computer-based investment programs also think every dip is a chance to buy.
The emergence of algorithm-based investment decisions based largely on the principle of following market momentum has to mean that the conventional process of price-discovery based on value and allocation of resources to the best social benefit has been compromised. This is almost a "driverless-car moment" for the banking sector - a move away from humans behind the steering wheel to a driverless, computer-based decision-making process that now dictates safety on our fast-moving investment highways.
Call me old-fashioned, but the emergence of algorithm-based investment decisions based largely on the principle of following market momentum has to mean that the conventional process of price-discovery based on value and allocation of resources to the best social benefit has been compromised.
This is almost a "driverless-car moment" for the banking sector - a move away from humans behind the steering wheel to a driverless, computer-based decision-making process that now dictates safety on our fast-moving investment highways. These programs are written with similar algorithms; thus, in unison, they reinforce the momentum of market movements.
One could argue that active investment management by fallible, greed-driven humans - which comes at a price for investors in the form of high fees and for taxpayers in the form of bailouts when things go wrong - is not sustainable. The economics of cost efficiency argue against it. A seemingly innocuous alternative is to have a machine make all the investment decisions, eliminating emotion, fear or favour. But is that the way to break the cycle of booms and busts?
Algorithm-based investment, the proponents argue, is a deliberate move away from having humans - who are susceptible to greed - make active investment decisions, and towards passive investment driven by cold, hard computer programs.
As with driverless cars, the benefit is that one mistake is a lesson for the whole fleet to learn from, not just for the individual driver or, in this case, the individual investor who is directly involved. But the risk is that small mistakes are ignored in favour of the existing momentum. By the nature of this trading style, major price swings can be exaggerated - hence the record highs currently seen in many stock indexes around the world.
REVERSAL NOT FAR OFF
Attentive readers would see the risk of this same momentum-driven investment style exacerbating the cycle when, not if, the reversal comes. Then, every uptick would be a chance to sell. Price corrections can be equally devastating and deep before the floor is found. The current exaggerated upswing in global stocks started in 2009 and, after eight years, in my view, the reversal cannot be too far away.
When the reversal does come, it might be like an entire fleet of driverless cars deciding to drive in reverse gear. Until the fleet hits the wall, nothing can stop it, and the momentum will drive it faster towards the wall, with potentially devastating effects - especially when the cushion provided by the vast amount of money arising from quantitative easing by central banks at the height of the financial crisis is simultaneously siphoned out.
The only difference between such a scenario and 2008 and other, earlier crises is that there would be no clear villain to blame. Rounding up the machines would be a Pyrrhic victory.
Indeed, the writing is on the wall. The US Federal Reserve is, in my view, set to siphon money from the financial system within the next three to six months, to normalise its balance sheet, and it will do so by not reinvesting maturing bonds in its holdings that were bought at the height of the 2009 quantitative easing programme.
The European Central Bank has begun to buy fewer bonds. The Bank of Japan is already buying fewer than it used to after having changed its policy targets in January. These decisions are driven not by algorithmic programming but by rising inflation - a macro phenomenon not captured by momentum-driven, algo-based investment strategies.
Worryingly, neither is the decision to siphon liquidity out of the system a factor ingrained in the algo-investment process. Admittedly, not even active investment driven by macro-nerds can predict fully what to expect when quantitative easing is reversed. No one has experienced that before, not even the Fed.
Without similar past experience, history offers little help to men or machines. Ominously, the best historical reference has to be the near-death experience of 2013 that was dubbed the "taper tantrum", when the Fed signalled its intention to reduce its bond-purchasing scheme so as to pump less liquidity into the banking system as the economy recovered. Then, emerging-market assets took a plunge as funds were withdrawn from seemingly higher-risk investment instruments. This soon proved temporary and was deemed, by the computer engineers, as a man-made folly driven by emotion.
In recent communications, the Fed has spoken about the withdrawal of liquidity. The same liquidity that has cushioned every form of market pullback in recent years is now being withdrawn. But no one - or more accurately, no machine - is worried. In the time between the 2013 taper tantrum and the end of last year, algo-driven investment funds have raked in more than US$2 trillion in fresh investors' funds globally.
Traditional macroeconomic- based fund managers will grudgingly admit that machines do matter - they are here to stay. But like me, a mere mortal economist, these seemingly immortal machines are, alas, fallible.
The writer is the Chief Forex and Rates Strategist at a foreign bank.
A version of this article appeared in the print edition of The Straits Times on April 08, 2017, with the headline 'Man or machine, both manufacture market bubbles'. Print Edition | Subscribe
We have been experiencing some problems with subscriber log-ins and apologise for the inconvenience caused. Until we resolve the issues, subscribers need not log in to access ST Digital articles. But a log-in is still required for our PDFs.