PsyFi Search

Saturday 15 May 2010

Complexity in Financial Systems

What's Complexity?

We can probably all agree that modern day financial systems are complex, but what that actually means isn’t something that everyone agrees on. Typically, though, a system characterised by complexity isn’t something that anyone’s designed – it emerges, it adapts spontaneously and it produces stunningly unexpected outcomes when no one’s expecting them.

Which, let’s face it, sounds a lot like modern finance. The problem is that many economists are focusing on how they manage this system when, in reality, it’s impossible to do so. It’s like trying to contain swine flu using a butterfly net.

Complexity Is Not Engineering

When many people, including economists, discuss complex systems they often use analogies with complicated engineered systems like aircraft or nuclear power systems. Now these are definitely complicated, with many, many interacting parts, the failure of any of which may compromise their integrity. However, complicated human engineered systems are not truly complex. For the most part the designers of these systems go out of their way to make sure that they don’t exhibit the trademark unpredictability and non-linear outcomes of complexity.

Indeed, the very fact that these systems have a designer is a sure sign that they’re not truly complex. This was the problem that Charles Darwin solved – how do complex things like human beings come into existence if not through the guidance of a designer? The answer, of course, is that complexity can arise through interaction with the environment as long as there’s some means of adaptation. These are the trademarks of complex adaptive systems. Darwin was concerned with biological evolution, but the global financial system is of the same type.

Complex Means Adaptive

There are two noteworthy things about complex adaptive systems. The first is that they’re complex. The second is that they’re adaptive.

And while that may be a statement of the bloody obvious it’s one that seems to escape many financiers studying the subject. The ability of the system to adapt, often in completely unpredictable ways, means that you can’t model it and you can’t foresee the outcomes of any strategy of intervention. It’s all completely unknowable in advance.

Once you accept this it becomes suddenly apparent that a huge swathe of modern finance is complete rubbish. For example, in a complex system you expect to see “tipping points” or phase transitions when the system suddenly and unpredictably switches from one stable state to another. As Caballero and Krishnamurthy have documented this appears to be exactly what happens during the episodes of liquidity hoarding and flights to quality associated with financial crises. People suddenly switch from a belief that they’re in a state where risk is measurable based on probability to one characterised by fear in the face of absolute uncertainty, so called Knightian uncertainty.

So, in the depths of the panic of 2008 we saw investors selling Collateralised Debt Obligations at almost any price largely because they didn’t know how to analyse them. What it looks like is that they bought these sub-prime backed securities because they’d been given the highest rating possible by the credit rating agencies. When some of these went bad the investors – many of them supposedly high powered institutions – belatedly recognised that they hadn’t got a clue about what they’d bought and sold, virtually at any price. One day they had nice risk models giving default probabilities, the next day they had junk.

The Theory of Rational Ignorance

In fact the problem is slightly worse than this. Many of these institutions had rationally decided not to invest in understanding the products they were investing in because they figured out that it was too expensive to employ the people capable of analysing the issues, an example of The Theory of Rational Ignorance. As Steven L. Schwarcz has remarked in Regulating Complexity in Financial Markets:
“The complexities of modern investment securities can lead to a failure of investing standards and financial-market practices for several reasons: these complexities impair disclosure; they obscure the ability of market participants to see and judge consequences; and they make financial markets more susceptible to financial contagion and also more susceptible to fraud.”
Our problem is that most of the time we’re not prepared to invest as though everything’s about to go wrong. We’re habitually biased to look at the upside, not the downside so we build portfolios without hedges, because they’ll restrict our short-term profits and we’ll always take the view we can get out before it all goes wrong. This is where an understanding of the instability of the global financial system can help investors: it spends its life permanently tip-toeing across a shaky wire over a gorge populated with extremely hungry crocodiles.

And it's got vertigo.

The Next Cause Will Be Different

There’s quite a lot of work now going into figuring out how we prevent the next crisis, much of it looking at how the financial system is re-engineered to prevent these types of failures. It’s a pointless exercise, because as fast as one set of rules is created the system will mutate to find ways around it. In so doing it’ll create all manner of unexpected connections, invisible to the overseers. Who, for instance could have predicted that badly designed mortgage schemes in middle America could have led to the collapse of the Icelandic government? Hidden links and invisible chains of consequence are everywhere in global finance.

Because the global financial system isn’t engineered and doesn’t have a designer there’s no way of controlling it. Those industrious academics trying to figure out how to use engineering concepts used to manage risk on complicated systems like aircraft are wasting their time. The failure of the global financial system isn’t like a single airplane unexpectedly crashing, it’s more like every single aircraft crashing simultaneously. It’s not the problems we can see, like the design of the avionics computer, that are the issue, it’s the fact that air-traffic control systems are dependent on a small number of global positioning satellites each of which happen to use the same specialised microprocessor, which has an unknown bug which is triggered by a combination of a certain date and freak sunspot activity.

Or something else, equally unpredictable.


Taken together the complexity of the global financial system, and the way that it’s often linked by tight but unappreciated couplings such that a small perturbation in one area can lead to devastating consequences elsewhere, makes it fundamentally unstable. Worse, however, is that even if you can get a handle on the risks today by tomorrow they’ll have changed. In such circumstances believing any prediction is an act of faith, rather than an intelligent assumption.

Equally any set of economic theories or financial models or loudmouthed gurus that claim to be able to foresee the future are hopelessly lost in their own rhetoric. There is no designer, there is no plan, there is no predictability. There’s just stuff, which happens. The best we can do either is plan with a margin of safety or just get lucky.

So, do you feel lucky, punks?

Simply Complexity: A Clear Guide to Complexity TheoryPatterns in the Dark: Understanding Risk and Financial Crisis with Complexity TheoryOrigin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics

Related Articles: Ambiguity Aversion: Investing Under Conditions Of Uncertainty, Panic!, Don't Overpay For Growth


  1. "There is no predictability..."
    I agree. But life isn't predictable as well. It's a probability game. And all events have near zero probability. That's why they are unpredictable. What was the probability last year this time that I would be siting here typing a response to your post. Indeed what was the probability 10 years ago that I would be living where I am and married to my wife?
    But...just because life is unpredictable doesn't mean it's uncontrollable. Everybody has choices in life that will put them in the gutter. Similarly our politicians make choices that can cause the financial system to break down or to work. Complexity doesn't mean, in my opinion, that all is lost.
    How about let's start with reigning in leverage and raising some capital requirements and even, yes, stopping the tax payer bailouts. I don't know if people have noticed but the financial system survived Lehman going under.

  2. Thank you for an excellent post. The complexity of systems is driven by two dimensions: variety in the behavior of each part and the level of connectivity/integration between the parts. The rise in the availability of real-time information has enabled players in the financial sector to increase the level of integration and connectivity, considering an increasing number of natural and man-made factors in their decision making: weather in the wheat-growing plains of Kansas, price of gold, politics in Greece. “Hungry crocodiles », as you said, and their computer program proxies still drive much of the way each part of the system responds (varies). This is one of key sources of variability in the system: How hungry crocodile perceive the future and react accordingly.

    This brings up a couple of additional questions: Does the way society evolves favor complexity? Is the financial eco-system bound to become more complex because of some driving force intrinsic to humans and nature?

  3. There’s quite a lot of work now going into figuring out how we prevent the next crisis, much of it looking at how the financial system is re-engineered to prevent these types of failures. It’s a pointless exercise, because as fast as one set of rules is created the system will mutate to find ways around it.

    I see this as the "There's Nothing New Under the Sun" view.

    My view is more a "History Doesn't Repeat But It Rhymes" view.

    There are certain basics of human nature that will cause things to repeat in a general way.

    But human progress is not entirely an illusion. We are capable of enhancing our understanding of things. So we are capable of reining in the power of those realities of human nature that otherwise would destroy us.

    Our old level of understanding is not today causing the same amount of damage as it has in the past. It is causing more. Why? Because we are a richer society. Investing ineffectively today is a more dangerous thing than it has been in earlier times.

    But that very reality may be the thing that pulls us safely to the other side! We need to solve this problem to survive. So the pressure will at some people get so great that we will permit consideration of solutions today viewed as too radical. "Poof!" goes our old understanding and all the troubles that came with it.

    I am a willfuloptimist!


  4. Tim, it seems to me there are some things that can be done to limit the damage of the next disaster, as another commenter mentioned, restricting leverage, stopping bailouts, also regulation/limits on derivatives, enforcing laws that may already exist, but are not well enforced. Your conclusion seems like it's saying nothing can be done, but it seems to me that most people with the ability to do something don't want to. Which isn't the same as can't. - Ilene (at - where I reprinted this article, thanks!)

  5. I'm definitely fatalistic about the trajectory of markets: they're adaptive and the nature of adaptive systems is to change abruptly, often in the ways that aren't to the advantage of the participants.

    So I do believe that markets will one day fail again and that there's nothing we can do to prevent this. However, I also believe that a lot more can be done to limit the damage caused by such failures: I wrote about this in Basel, Faulty? Currently we try to train the tightrope walker not to fall off, whilst at the same time incentivising them to get to the other side faster. This is wrong, we'd be better off constructing some safety nets under the tightrope and accepting a little less danger in our lives.

    In response to c0mplexity's questions on whether it's inevitable that complexity will increase - well, the evidence from biology suggests not; complexity can decrease under the right environmental pressures. Much complexity in financial markets arises in order to confuse - as we saw in Finance: Where the Law of One Price Doesn't Apply a few basic financial products are continually repackaged to avoid the profit erosion caused by allowing consumers to make pricing comparisons. Reducing complexity there would be a good start in creating a safer world.

  6. Thanks for addressing my question, Tim.

  7. My experience with this kind of complexity comes from working with international taxation of financial activities. It doesn't look pretty.

    Changing the rules does help, but is of course no permanent fix. The constant increase in complexity to a system wears it down, just as we are familiar with in the physical world. Use your car, and you get closer to next time you have to get new tyres. The wear on a set of rules in the form of increased complexity (usually because someone design new facts in order to exploit some weak spot in a specific rule) is a function of time and the intensity of the use of the specific rules. After a certain time and use you have to change the rules again, or you will have a nasty accident.

    Goes for tyres as well as tax code.