Skip to content

Banks and Biology – the Super-spreader Analogy


What do the financial system and a tropical rainforest have in common? According to Andrew Haldane, executive director for financial stability at the Bank of England, a lot. Both are complex adaptive systems and comparing them would allow to identify and analyse financial structural vulnerabilities and draw important conclusions about how to best manage the financial network.

The idea has been elaborated in various forms by Haldane (2009)  and Haldane and May (2011), (THE Lord Robert May, professor of zoology at Oxford University and former British scientific advisor), and the Economist mentions a forthcoming paper for Nature. As fascinating as it is, the idea may contribute less to the recent debate than it appears at first glance.

The line of reasoning is roughly as follows:

–          These days, the international financial system is characterized by two features: complexity and homogeneity.

–          The combination of the two would have set alarm bells ringing in non-financial disciplines such as ecology or engineering, because in those disciplines together they stand for fragility and instability.

–          In complex systems scaling up risks – as is understood to be done, for instance, by portfolio diversification under the assumption that on average they will cancel each other out – may result in building up “error cascades”. The reason is “cross-contamination”. The system is mostly self-repairing but exhibits a knife-edge property which under growing homogenisation and complexity bears the danger of collapse.

–          Accordingly, in complex systems, the focus should be more on diversity (in contrast to homogeneity) than on diversification which may make banks move in the same direction at the same time thereby aggravating imbalances.

–          Banks are connected to many others making them ideal candidates for risk contamination. The biggest, most complex and best connected ones are “super-spreaders” with a high capacity to infect counterparties.

–          Instead of focusing on risk related to institutions policy should concentrate on system-wide risk and on financial networks.

–          Forcing few big banks to adopt safer practices could significantly reduce system risk. Here, the soundness of institutes, or lack of, is no criterion. What matters is size, complexity and numbers of market counterparties.

–          Banks that are identified as “super-spreaders” should hold higher amounts of loss-absorbing capital.

Appealing as it looks the approach leaves many open questions.


What is complex in this context? Is it banks? Is it activities or financial instruments? Or is it markets and the financial network connecting banks?

Actually, in Haldane (2009) a bit of all of them can be found. There, system complexity is determined by the network of nodes and the number of links between them. Empirically, in the paper nodes are country data of the sum of cross-border stocks of external assets and liabilities of 18 countries, and links are bilateral external financial stocks relative to GDP. In a world of almost perfectly footloose finance, high market concentration and widespread electronic trading this must appear a somewhat strange approach since probably the same handful of “super-spreaders” is trading and doing business in each of those places. On the other hand, appropriate bank data are presumably not available.

The difference matters reflecting various stages of world financial relations as the following figures try to illustrate. The first one demonstrates traditional bilateral links between countries in the early days of finance with financial services largely following, and adjustings to the needs of, trade:

At the next stage, third-country foreign activities grow in the process of internationalisation, and increasingly banks are doing business on their own independent of trade:

The charts in Haldane (2009) provide a fascinating account of this development allowing to trace the process of European financial integration, the growing interest of investors in the emerging economies and growth markets of Asia and Latin America in the 1990s and the rise of new world financial centers outside Europe and the US.

At the third stage, the distinction between home countries of financial institutions has been abandoned. Banks have become truly global with regional centres in all major places in the world, as well as some smaller ones, doing a large part of business with one another. Talking about financial networks, this is the scenario policy is confronted with:

According to Haldane, complexity is also found on the product side. Here, the emphasis is on the role of derivatives such as CDSs and of securitisation and structured financial instruments such as CDOs in recent years. The latter generate chains of claims thereby amplifying uncertainties about counterparty exposures and making related risks “almost unknowable”.

To briefly illustrate the point: A Collateralised Debt Obligation (CDO) is characterised by three features, pooling, tranching and de-linking. Cash-based or synthetically created assets are bundled or put together (pooling), and with the help of computer modelling cash flows from this pool are then splitted into separate classes of liabilities with different risk and return characteristics (tranching).  De-linking means that the credit risk of the asset pool is separated from the originating firm through use of a special purpose vehicle (SPV). Structures can become very complicated. Haldane gives an example where an investor would have to read 1,125,000,300 pages of documentation to fully understand the ingredients of a product. The following figure shows the basic principle:

Finally, banks are complex, too. Haldane names Lehman Brothers as example. When filing for Chapter 11 bankruptcy the firm was believed to be counterparty to around $5 trillion of CDS contracts.

Large numbers

This example demonstrates the ambiguity of the concept leaving ample scope for misunderstandings. A couple of years ago, complexity has been one of the watchwords of the scientific community – beside “self-organisation”, “chaos” and “dissipative structures” (Coveney and Highfield 1990). Robert May, Haldane’s co-author, is well-known for his research on population dynamics. Traditionally, irregular fluctuations of natural populations were explained by external random influences such as climate and disease (Huggett 1998). May demonstrated that these could result from intrinsic non-linearities in the dynamics.

In the literature of chaos and self-organization complexity is a characteristic of systems, not of their individual parts. These systems can be understood only from a holistic view rather than a description of the parts (Bak 1997). Whether a single grain in a sandpile  – to take an example from Bak 1997 – has a simple or complex structure is only of interest if this individual characteristic affects its interaction with other grains in the sandpile, i.e. if it adds something that makes the whole of the sandpile become more than its parts. Complexity is about interaction as Haldane, too, emphasizes.

Applied to financial systems, complexity would arise from the interaction of banks and other financial and nonfinancial actors in the system. In this respect, it does not matter – to come back to the above example – whether Lehman Brothers was counterparty to trillions of CDS contracts or only to 5.000. What matters is with whom  Lehman was doing this business, in which way and in which environment. There is no way to understand a complex financial system by focussing on an individual institution.

In complex systems interaction must take a specific form – model parameters and initial conditions must be “right” – for random-looking behaviour or “collapse” to emerge. In the literature, the mechanisms behind the dynamics have been described as “stretching and folding”: Periodic patterns of different length arise and are overlapping, influences enhancing the system interact with retarding ones. In a tropical rainforest it is the interaction of multi-layered canopies in the competition for light and growing space. In economies, patterns that overlap can be found in business cycles, the rhythm of the seasons – construction in summer, Christmas sales in winter -, the ups and downs of stock markets, high-frequency movements in financial data and many more.

However, the interplay of these forces does not necessarily produce the kind of excesses which in Haldane’s words risk rendering the system “non-renewable”. The range of possible outcomes includes periodic and quasi-periodic behaviour which to an outside observer may appear quite dramatic as well. Ian Stewart gave an example of the interplay of forces in a quasi-periodic system which may help illustrate the point:

Imagine an astronaut in a lunar orbit swinging a cat around his head in a space capsule … The cat goes periodically round the astronaut, the astronaut goes periodically round the Moon, the Moon goes round the Earth, the Earth round the Sun, and the Sun revolves round the centre of the galaxy. That’s five superimposed periodic motions.” (Stewart 1990: 104)

Not knife-edge criticality, but long-term invariance. Self-healing, self-repairing it might be called …, almost always converging to a hidden structure no matter what the initial conditions and how chaotic its behaviour may look to an outside observer.

Does the financial system work this way? We don’t know and probably, there is no way to empirically find out.  But, lack of knowledge calls for further research, not for policy prescription.

Losses and fear

Gillian Tett rightly noted that from the attack on the World Trade Center, the collapse of Enron and the implosion of the internet bubble the system had bounced back fast. Financial crises seem to come and go with remarkable “quasi-periodicity” and even if the one that began in 2007 appears frightening in its scale and consequences there is no reason to believe that in its dynamics it differs fundamentally from earlier ones.

Nevertheless, the similarities between nature and finance appear striking, the image of spreading disease is a forceful one and there are lessons to be learned from comparing the two:

Following Haldane (2009), let the process start in the two systems with an external shock, an outside event. Examples are the emergence of SARS on the one hand and a market failure on the other. The immediate results are illness and losses, and in both systems contamination (more sick people, more losses) occurs. Similarities are also found in the reactions to the event which in both cases is fear. But, at this point the analogy shows first cracks because this fear has different consequences: In the SARS case, there are secondary effects on the economy, but fear has no influence on the spread of the illness. In the financial system, beside direct contamination – losses from payment delays or default – fear is the driving force behind crisis dynamics.

Another difference with considerable implications concerns the parts of the system and its “topology”. While in biology, the system consists of a homogeneous population where diseases spread more or less equally, the financial system is a rather heterogenous conglomerate of banks, shadow banks and non-banks.

Eco-system Financial system
System parts Population Banks, shadow banks, non-banks
System topology Homogeneous Heterogeneous, hierarchical
Network size Large Small
Network characteristic Transparent Intransparent
Trigger External shock (SARS) External shock (market failure)
Result Illness Losses
System effects of result Contamination Contamination
Reaction Fear Fear
Effects of reaction Secondary economic effects Contamination
Policy target Restoring health, saving population Restoring health, saving system components (credit business)

As a consequence network characteristics in both systems differ as well. In biology, the network is large and populations are atomistic. Epidemiology is a science of the many. In finance, the network is much smaller. There is a two-tier structure of a number of big banks at the core and smaller ones at the periphery. The core network is an oligopoly doing business both with one another and with clients. This network exerts a dominant influence in many areas of finance. Its representatives are in constant dialogue with regulators and policy.

In addition, there are so-called shadow banks, providers of financial services such as taking deposits and making loans that take place outside the regulated system. Examples are investment banks, hedge funds, money market funds and insurers:

Error cascades

As a consequence of this structure other network characteristics differ, too. In biology, as a rule, transparency prevails. Sources and spread of infections are easily traced. In contrast, financial relations are opaque, and in particular official regulation offers incentives for this state to prevail. One example is securitisation which was “invented” in reaction to Basel II with the intent to remove assets from balance sheets thereby reducing capital charges.

Innovations generally bear the risk of becoming a hype. In this, CDOs do not differ fundamentally from the bulbs of the recently introduced tulip in the 17th century or the shares of the newly established South Sea company in the 18th century. Bubbles emerge, grow and burst. As irrationality spreads, so does “Knightian uncertainty” (Haldane) in pricing assets, but neither investors nor regulators seem to care before it is too late.

One instrument to at least partly shelter financial institutions from market excesses is portfolio diversification. The idea is to invest in different markets and instruments in order to reduce overall investment risk. In not putting all eggs in one basket, so the proverb, but carrying them in several baskets the risk is reduced to lose them all if a basket is dropped. Increasing the number of eggs in a basket, as Haldane rightly puts it, increases risk. But this is not the risk of adding a bad egg. A bad egg – or as in Haldane and May (2011) a bad apple that threatens to contaminate the whole basket – would be a problem which could easily be solved by close monitoring: Simply find the apple and take it out!

The idea of diversification has been misused by CDO issuers and investors. Piling up huge positions in individual instruments in search of returns without proper means to evaluate risks clearly hurts the principle. Portfolio diversification does not offer perfect protection at the burst of the bubble when correlations of markets converge and trading even comes to a halt. But even in these cases it prevents banks from being among the biggest losers.

Policy response

Homogeneity, once identified, sets alarm bells ringing not only in the natural sciences but among economists, too, as it signals the danger of speculative runs and bubbles, independent of system complexity. However, with respect to official reaction, again, there is a difference between biology and finance.

In biology, the main target is to restore health and save the population. In finance, the main policy interest is to preserve important systemic functions such as credit provision. The aim is not to save the patient by all means. Usually, a distinction is made between solvency and liquidity. One immediate policy reaction to crises is to avoid liquidity shortages. In cases of insolvency, the rule is to let the patient die. The big challenge then, of course, is the “too big to fail” case.

As the Japanese experience demonstrates, in principle, to let big banks go bankrupt is possible,  but it requires very special circumstances: When in the 1990s in Japan the property bubble burst and the system became overloaded with bad loans,  the Japanese government, like the US authorities in  2007, failed to tackle the crisis at an early stage. But almost unnoticed by the world, in the following years it undramatically managed to fundamentally restructure the whole system orchestrating mergers and acquisitions, closing banks and – very unusual for Japan – even selling one of the biggest, LTCB, to a consortium of foreign banks led by Ripplewood Holdings. All this happened with remarkably few disruptions. There were no signs of widespread fear or panic that the system would collapse.

This phenomenon could not be explained by Japan’s formal rules and regulations to cope with the crisis which were wholly insufficient (and, as I wrote elsewhere, we still do not fully know the price that has been paid for this). The reason why the system continued to work was a widespread trust of the public, not necessarily in individual politicians or parties, but in the system and in authorities and official institutions in charge which prevented a wider contamination and fear.

Trust is the key for providing authorities with the scope to deal with the “too big to fail” problem and where necessary to let even big banks go bust. Not banks’ networks, not the “super-spreaders”, but their countless bank and nonbank clients and the wider public should be the first adressees of policy in rewriting the rules of the game. Whenever the system would reach a critical state, the trust or fear of these people would decide about its survival and not network dynamics. Increasing transparency, bringing shadow banks under the roof of official regulation and redefining the rules for products and trading practices – focusing on systemic risk cannot spare regulators from searching full understanding of products and legal constructions – as well as clear signals of determinedness would help build up confidence.

As Howard Davies wrote in his book about the causes of the financial crisis: In the current situation, the problem is that policy responses are being proposed, and even implemented, based on narratives which may not be well supported by the evidence.

The weakness of the narrative of complex adaptive financial networks is that there is no way to find out empirically about the system’s true nature. If policy feels more comfortable with big banks holding an additional capital cushion, be it so. But the super-spreader analogy could offer no justification for this approach, and if the next crisis strikes it will not prevent panic and runs and markets drying up bringing the system to the brink of collapse once again.

References and materials

The Financial Crisis Inquiry Commission (FCIC) has compiled a library of selected CDO documents.

Howard Davies (2010) offers a very lucid “opinionated guide to the arguments about who or what was to blame for the crisis”.

Additional introductory material can be found on the Wiley Companion Website of Reszat 2005, Lecture 4, Case Study: Structured Finance.

From → Markets

One Comment
  1. I am enjoying these posts. I am a big fan of May’s approach to bridging ecology and finance. I am particularly interested in the tension between individual versus systemic risk that you allude to because it so closely follows so many other classic social dilemmas.

    Just a minor note of where I disagree with you. You write that “In biology, as a rule, transparency prevails.” From my experience, though, ecologists actually think the exact opposite. Ecological data on food-webs, interaction strength, populations, etc is considered to be much much harder to gather than the analogous data in finance. As such, I was under the impression that a lot of ecologists were looking to finance so that they would have access to better and bigger datasets with which to test their ecological theories. In particular, I think ecologists would be extremely happy to have biological analogues of data sets like these:

    Huang, X., Vodenska, I., Havlin, S., & Stanley, H. E. (2013). Cascading Failures in Bi-partite Graphs: Model for Systemic Risk Propagation. Scientific reports, 3.

    Soramäki, K., Bech, M. L., Arnold, J., Glass, R. J., & Beyeler, W. E. (2007). The topology of interbank payment flows. Physica A: Statistical Mechanics and its Applications, 379(1), 317-333.

    The part where I think ecology and finance really differ is in that the scientists can imagine themselves and their work as outside of the system they study. However, a scientist working in finance has to account for the fact that her recommendations or observations will be absorbed by the market and potentially affect its behavior.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: