Risk Modelling: Is the Response to the New Complexity Really So Simple?Published: October 16, 2012 in Knowledge@Australian School of Business
The unstable global economy poses serious challenges to both companies and pension funds. Nowhere has this been more obvious than in the area of risk management.
Rapidly changing global conditions have been particularly disruptive to the more precise risk management methodologies that enjoyed the high watermark of confidence levels during the period of macro-stability after World War II. At that time, extraordinarily stable conditions provided an environment that lent itself to mathematical applications and the ability to forecast with higher degrees of confidence, notes Andries Terblanche, a professor at the Australian School of Business and chair of financial services at Big Four consulting group, KPMG.
Of course, the world has changed. Significant among the financial and economic changes have been a big increase in the number of floating currencies, a significant growth in the use of derivatives, higher levels of public debt and an increase in global capital flows, aided by new technology.
Post-1970, more than 15 globally significant economic meltdowns have occurred, but not one was satisfactorily predicted by mainstream financial models, observes Terblanche. Yet risk continues to be calculated mathematically with assumed levels of confidence. "The expectation that the maths will tell us everything we need to know is simply no longer valid," he says.
Terblanche suggests that volatility assumptions and correlations used in risk models should be adjusted more frequently going forward. His views are consistent with those of Andy Haldane, executive director for financial stability at the Bank of England.
In August, Haldane made headlines around the world with a riveting speech at the Federal Reserve Bank of Kansas City's 36th economic policy symposium in Jackson Hole, Wyoming. He told a group of central bankers that the rules governing risk in financial institutions have become so complicated they should be thrown away. He called for regulators to go back to the drawing board.
"We're complicating matters when the right way forward is to keep things simple," says Haldane, warning against layering complexity upon an already too elaborate set of rules. What should be happening is the opposite. The regulatory solutions being planned so far won't prevent another financial crisis. "Applying complex rules in a complex environment is a recipe for catastrophe," he stresses. A rules-focused approach is potentially compounding the error when there should be a greater reliance on judgment.
Haldane's speech was titled The Dog and the Frisbee and it has sparked a new way of looking at risk - or, arguably, a return to an old approach. Haldane uses the analogy of a dog catching a frisbee to make his point. Dogs don't stop to consider the aerodynamics that make a frisbee fly, such as wind speed, or air resistance; they just instinctively know to run at a speed that keeps their angle of gaze to the frisbee constant. Baseball players and cricketers use the same rule of thumb.
"To ask today's regulators to save us from tomorrow's crisis using yesterday's toolbox is to ask a border collie to catch a frisbee by first applying Newton's law of gravity," says Haldane. The solution is to focus instead on the root causes of financial disasters. The more this view is obscured through voluminous legislation, the more likely that mistakes will be repeated.
Terblanche similarly believes that obsessing about detail is not the solution. "The more we obsess about technicalities and mathematise solutions, the more we run the risk of missing the obvious - something that could adversely affect the very objective of our risk modelling," he says.
The Human Factor
Haldane is a strong advocate for a return to behavioural finance, a lesser-known, controversial branch of academic thought that took hold in the 1990s. Behavioural finance theory purports that people do not behave rationally but are swayed by psychological and sociological factors. It undercuts the idea that capital markets are efficient and spurns the view that future outcomes can be predicted in a highly complex and interconnected world where the number of risks is infinite.
Haldane points out that the largest super-computers cannot compute many more than 10 moves in advance. Chess grandmasters are unable to evaluate fully more than five chess moves ahead. "Yet most real-world decision-making is far more complex than chess - there are more moving pieces with larger numbers of opponents evaluating many more moves ahead," he says.
Risk is conventionally taken to refer to an exhaustive range of outcomes for which probabilities are defined. However, Haldane maintains that the uncertainty now facing the world presents different circumstances. With uncertainty, the range of outcomes may not be known, let alone the probability of them occurring. "Defining future states of the world, and probability-weighting them, is beyond anyone's ability," he says.
With risk, says Haldane, policy should respond to every raindrop; it is fine-tuned. With uncertainty, the response is to every thunderstorm; it is coarse-tuned and simple decision rules are best.
He cites a number of instances where people have sought simple behavioural rules to cope with a complex environment. Doctors diagnosing heart attacks find simple decision trees beat complex models. Detectives tracking down serial criminals find that simple locational rules trump complex psychological profiling. Investors have found simple passive strategies outperform complex active ones.
A Big Call
Haldane wants to change the way we think about economic science and it's a pretty big call, says Chris Adam, associate dean and a professor of finance at the Australian School of Business. "It just sweeps away most of what's in the textbooks."
Adam notes that behavioural finance "doesn't provide us with clear predictions of events. At most, it is very good at looking back and working out why we get things wrong and at identifying and analysing outcomes that don't fit classical finance theory." It helps to identify that people made the wrong decisions or they focused on particular points when they should have been looking more widely. "Rather than estimating a currency exposure and hedging it with a probabilistic pricing model, behavioural finance can run a range of models and suggest myriad possible actions," Adam says.
Still, behavioural finance has a novel aspect in moving finance into an experimental area and Adam advises running controlled experiments comparing how people make decisions using both classical finance and behavioural finance to see which gives the most useful answer.
The problem is, behavioural finance can't predict: it can find solutions to real world problems using genetic algorithms and the process of evolution. Using computer-generated codes, it can create generations of agents with particular behavioural characteristics which interbreed. "You can get unusual outcomes which a standard economic model can't explain ... such as the emergence of altruism," Adam says.
Due to the inability to predict outcomes, behavioural economics models cannot be used in the same way as the classic models, such as in predicting tomorrow's exchange rates.
"The danger of taking an experimental approach is that it can sweep away a lot and may not leave much in its place," cautions Adam. "We're fishing in very deep water in making these comparisons. A sweeping reassessment of classical finance and economics is neither feasible nor desirable."
So, how should the new uncertainty be approached? Terblanche suggests there will always be a role for behavioural finance. "However rational we think we are, we disappoint on rationality measures," he says.
As a result, it's hard to provide a single, comprehensive model that can encapsulate human complexity and, at the same time, provide certainty in its outcome. "The more you obsess about arriving at a single solution that encompasses human behaviour, the more you run the risk of missing macro-trends and changes which, individually and collectively, can produce new risks," says Terblanche.
His response is to not throw the baby out with the bathwater - and to retain the development in risk management that we have. Importantly, the limitations of existing models, including the assumptions, need to be understood.
Risk modelling needs to be expanded to incorporate macro shifts in our operating landscape, which may appear insurmountable in light of the sheer volume of information. There are established mechanisms and models to distill the information to key issues that matter, significantly, to businesses. "Companies should therefore focus on macro-economic, socio-political and other large-scale changes, their interconnectedness and their velocity of disruption in order to identify upside and downside," says Terblanche. Leading organisations are already doing this. Many central banks have also seen the light and have established prudential stability units or departments.
A recently released survey carried out by Natixis Global Asset Management, one of the world's top 15 asset managers, reveals that managers have completely backed off the accepted financial models and investment theories in the belief that they will fail.
More than two-thirds of Asian institutional investors surveyed are convinced the diversified portfolio approach is no longer the best way to pursue a steady stream of income over time, and are focusing on allocation of risk rather than capital, which is a startlingly different approach.
Changing Rules to Counter Risk
The Bank for International Settlements says none of the problems that contributed to the global financial crisis have been comprehensively resolved. Haldane points out that risk models are now too complex. The Basel regulatory framework, which started at 30 pages in the late 1980s, will run to some 60,000 pages when Basel III is finalised. So why do it?
For Terblanche, the answer is, in essence, straightforward. "When ordinary investors lose money on a large scale, governments feel a sense of duty to respond. The bigger the market adjustment, the bigger the response in the attempt to ensure it does not happen again," he says.
Justin O'Brien, a professor in law at the University of New South Wales and director of the Centre for Law, Markets and Regulation, approves of the call for more effective supervision rather than a reliance on rules. He says that equating effective risk management with compliance is a fundamental misconception: "It's important to change the culture of regulations, not rewrite the rule book. We need to rethink whether or not we are creating the right kind of regulatory environment."
O'Brien suggests that rather than looking at "how we regulate, we all need to step backwards, review the entity we're supervising, and ask what we're regulating and for what purpose".
"[Haldane's speech] has come at a very interesting time. If you look at the recent Treasury select committee report into LIBOR, it points to a fundamental problem, not just with the culture within Barclays but banking as a whole and with oversight. A complicated system of regulation creates circumstances whereby those rules can be transacted around. And, the more complicated the structure, the more invasive the oversight has to be," says O'Brien.