Model Thinking is Flawed

Relying on models can be disastrous.

Bullion.Directory precious metals analysis 31 March, 2015
By Christopher Lemieux
Senior Analyst at Bullion.Directory; Senior FX and Commodities Analyst at FX Analytics

Financial modeling has been heavily relied on by investment firms and banks for decades, but the these models have acute limitations. These models are overly complex mathematical representations of a particular financial situation and how it is affected by a series of outcomes. 

Over reliance on this model thinking has caused a great deal of risk to financial institutions who tend to get blind-sided by outside factors deemed not plausible or risks that were not even factored into the model itself. The result tends to be a nice and neat, mostly optimistic, view of what could happen in the future. 

For instance, Goldman Sachs executives commented on the Swiss National Bank’s decision to un-peg the Swiss franc from the euro, which caught market participants off guard. The result led to a 41 percent appreciation of the central bank suppressed franc in the matter of minutes and a 15 percent crash in the Swiss equity market. According to the executives, the volatility that ensued was almost unheard of in terms of their models. The pair sank 3,000 pips in virtually a blink of an eye.

Although in hindsight, largely because I do not follow the EURCHF closely, I noticed the daily volatility (translated in pips) begin to widen, slowly at first going out as much as 10 weeks before the SNB announcement. In January, the daily range grew larger, and the week prior to the announcement it was multiples of what it used to be back in November.

The result of the SNB’s decision are well-known, causing millions of traders to lose millions of dollars. From retail trader to investment banks, like Citi, nobody was spared and the carnage was grand in nature. It was an event that, according to those with P/Ls ripped apart, defied all odds. However, those prudent knew that the SNB could not buy euros forever. Just the sheer size of the central bank’s balance sheet could have indicated the end-point, reaching nearly 90 percent of Swiss GDP.

Jim Grant, The Grant Interest Rate Observer, noted in his September issue that there was a likelihood that the peg would end, while even offering a potential trade idea that really paid off in January. Of course, it was unknown exactly when this peg would come undone, but nobody seemed to think it was even possible because their models did not suggest it was possible.

Mark Branson, director of the Swiss Financial Market Supervisory Authority, said that this shows “just how much faith the banking sector places in models and their predictive power.”

The problem is that models can be manipulated and often the result of one’s systemic thinking. If the model never included the potential risk of this magnitude then it is unlikely that those relying on these models would even have any defense to hedge against it.

Much of the of 2008/9 financial meltdown was contributed to the fact that these complex models failed and were quite flawed. Paul Wilmott, founder of the Certificate in Quantitative Finance, believes the models that were suppose to define risk actually exacerbated the fallout.

A lot of people who worked in derivatives actually thought they [models] were actually much better then they were. And there is an incentive built into the system, a kind of moral hazard, that encourages people to believe in these things because it allows them to trade, and trade bigger, and bigger and bigger. And, a lot of problems with risk management techniques, for example, is that they can be used to hide risk when they should really be alerting you to risks.

Wilmott also has disused that models were too flawed and too complex because there was no real application in the real world. He also warned, prior to the meltdown, that these derivatives would be at the very root of what we now know as the Great Recession, in the US, but the affects were felt worldwide.

In the short-film Qunats: The Alchemists of Wall Street, Wilmott was briefly explaining non-linearity through an example of purchasing beer for a party. During the example, he said, ‘and that’s the real world, but that’s not in 99.9 percent of finance models.”

Economic models are no different. The Federal Reserve is full of very intelligent people. I have no doubt Ben Bernanke or Janet Yellen are very intelligent. Yet, how can those holding Ph.D’s in economics be just horrid forecasters? They rely greatly on these models to predict outcomes, but the very nature of economics is to study human behavior within an economy. It’s dynamic, but model thinking is not the way it is currently used.

The Fed thought quantitative easing would boost aggregate demand, and it did not. Instead of consuming more, debtors began to pay down debts and save. Creditors took less risks and lending came to a standstill. This was unfathomable to the Fed.

Their model thinking led them to believe QE would be a cure all to the vast economic aliments. Their thought process is so narrowly defined, when QE1 failed, they began QE2, when that failed it was Operation Twist. That failed, and it was on to QE3. There is strong debate that after $4.3 trillion in Fed balance sheet expansion, the economy is no better off then it was previously.

To sum it up, Wilmott on the divergence of economics, models and the inability for real-world application:

… I also think there is a problem with economics. I think of economics as sort of a disorder on the autistic spectrum, if you like. What I mean by that is, people on the autistic spectrum is they don’t have an understanding, an intuitive understanding of how the world works, which is what we very see with economists; and people on the autistic spectrum, or as to those they call NTs or neuro-typicals, they create their own internal models of how the world works… I don’t think economists see things, they don’t look around them. They don’t get the experience from the real world.

Models need to be less complex and more applicable to the dynamics of the real world.

 Previous Post

Centrally Planned Markets Move Higher on China QE Hopes

Next Post  

CERN Scientists Create $230m Gold