The world economy was on the brink of collapse in the autumn of 2008 following the failure of Lehman Brothers. Confidence – the lifeblood of the financial system – was evaporating at an alarming rate, financial institutions refused to do business with each other, people took their money out of banks, and it looked like the economy might be heading for a second Great Depression. Then, just as suddenly as the crisis materialised, it seemed like it was over – while the economic, social and political repercussions continue to this day, the immediate threat of disaster is gone.
What happened in 2008 was a near-miss systemic crisis. Systemic crises are generally defined as states in which interlinkages and feedback loops within the financial system lead it to stop performing its essential roles. This event was followed by an economic depression and a sovereign debt crisis in the Eurozone, which in turn worsened the global financial crisis. A full collapse of the entire financial system was only averted thanks to the swift actions of the policy authorities – the central banks and finance ministries of the leading economies – which stepped in to bridge some of the defects in the system.
But this was not the first time that the world has faced systemic risk. The threat and, at times, its realisation have been present ever since the first financial system was created. They are an inevitable part of any market-based economy (see Danielsson 2013).
Systemic risk and the Global Crisis
So what led to the most recent build-up of systemic risk and the resulting Global Crisis, an outcome that seemed to catch almost everybody by surprise? One explanation is that the relative absence of crises over the past few decades lulled policymakers, financial institutions and researchers into complacency. Central bankers mostly focused on fighting inflation while regulators tended to neglect the system as a whole, instead aiming to ensure that each individual financial institution was well regulated – through so-called prudential regulations.
The latest crisis has demonstrated the folly of such thinking, which assumes that as long as each component of the system is safe, the whole system must be safe as well. This assumption carries the danger of policy authorities falling victim to the ‘fallacy of composition’ prefigured by Milton Friedman (1980):
“The great mistake everyone makes is to confuse what is true for the individual with what is true for society as a whole. This is the most fascinating thing about economics. In a way, economics is the most trivial subject in the world, and yet it is so hard for people to understand.
“Why? I believe a major reason is because almost any interesting economic problem has the following characteristic: what is true for the individual is the opposite of what is true for everybody together”.
With financial market regulation, the fallacy of composition has serious consequences – attempting to ensure the safety of each part of the financial system independently can lead, perversely, to the system as a whole becoming more unstable. Trying to make every market participant behave prudently destabilises the financial system.
One mechanism that allows this to happen is shown in Figure 1, which illustrates how a regulatory requirement for all banks to be run prudently can create systemic risk if it prompts the sale of risky securities in response to an external shock. If all banks respond in this way, a downwards spiral of asset prices is created, forcing banks into more sales and further depressing prices, resulting in a crisis.
Figure 1. When individual prudence leads to a systemic crisis
Attempts to reduce the risks related to any one institution and to dampen the natural volatility of the markets over the short term lead to a ‘Great Moderation’ – a seemingly permanent state of stability. But this false sense of security is itself the cause of a hidden build-up of systemic imbalances. The outcome is a manifestation, perhaps in a novel manner, of Hyman Minsky's famous dictum – ‘stability is destabilising’ (Minsky 1982).
This is the key insight of the emerging field of systemic risk – the threat comes not from outside the financial system and the economic, social, political and legal setting in which it is embedded; rather, it comes from interactions within the system, possibly amplified by the structure of the system and the ‘rules of the game’ established by the policy authorities at national and international level.
This way of thinking – in terms of endogenous amplifying effects in both the build-up and unfolding of a systemic event – is applicable to real world situations, past, present and future, and forms the basis of systemic risk modelling (Systemic Risk Centre 2015).
Systemic risk and endogenous risk
The Financial Stability Board, the international body created in 2009 to oversee the global financial system, defines systemic risk as follows:
“The disruption to the flow of financial services that is (i) caused by an impairment of all or parts of the financial system; and (ii) has the potential to have serious negative consequences for the real economy”.
There is broad acceptance of this as a descriptive definition but a more fundamental understanding of systemic risk requires a deeper sense of what constitutes a ‘system’. After all, how can one think about systemic risk without having a clear sense of what a system is in the first place?
Zigrand (2014) defines a system as a functioning mechanism governing a set of elements:
- That makes reference to something or to a central concept in a coherent fashion (‘deductibility’);
- That implies meaningful relationships between its elements (‘irreducibility’); and
- To the extent that it evolves, that it keeps its identity.
Examples include a ‘banking system’ (with a central bank at the centre of a network of interrelated banks) as opposed to a collection of banks, a ‘payment system’ as opposed to a set of bilateral payment and settlement arrangements, a ‘solar system’ as opposed to a cluster of celestial bodies, and a ‘nervous system’ as opposed to a collection of unrelated nervous cells.
The term systemic risk comprises the risk to the proper functioning of the system as well as the risk created by the system itself. The risk that is created or amplified within the system is ‘endogenous risk’. In the extreme, the risk may be a risk to the very central concept that guarantees the logical coherence of the system in pursuit of the best use of scarce resources with multiple ends.
Instances during which the central concept is itself incapacitated in the four systems outlined above could involve the following:
- The inability or unwillingness of a central bank to act as a lender and market-maker of last resort removes the foundations to a banking system;
- A failure of the real-time gross settlement processor in a payment system brings the system and all connected systems to meltdown;
- A stroke incapacitates a nervous system; and
- Hyperinflation reduces a price system to a set of primitive and inefficient bilateral barter operations.
In each case, the system stops being able to fulfil its function properly and consistently in a systemic event.
Standard methodology for modelling risks treats systemic risks as being mainly extreme shocks from outside the system drawn from some distribution – for example, the payment system would grind to a halt in the event that an asteroid hit the processor. But while the asteroid would certainly constitute a risk to the system, it is more fruitful to focus on endogenous risk, where the risk is both to the system and amplified by the system. One example is from John Maynard Keynes (1936):
“By a cyclical movement we mean that as the system progresses in, e.g. the upward direction, the forces propelling it upwards at first gather force and have a cumulative effect on one another but gradually lose their strength until at a certain point they tend to be replaced by forces operating in the opposite direction; which in turn gather force for a time and accentuate one another, until they too, having reached their maximum development, wane and give place to their opposite”.
This is the approach we adopt – to view systemic events as being mainly endogenous, with the trigger for a crisis possibly being exogenous though not by itself extreme. On that basis, systemic risk may be defined as the risk of a systemic event occurring, where a systemic event is defined by the occurrence of positive feedback loops within the given system that adversely affect the proper functioning, the stability and, in extreme cases, the structure of the overall system itself, with resulting costs to the wider real economy of which the system is a subcomponent.
Feedback loops and amplification mechanisms
Within the financial system, a small event can turn into a major crisis – a systemic event – while a much larger shock may whimper out into nothing. Behind those shocks that become systemic events is the presence of mechanisms that amplify and/or accelerate the impact through the rest of the financial system. Amplification mechanisms are the ways in which endogenous risk manifests itself in the financial system and translates into concrete events.
There are a number of features inherent in the financial system that can amplify a small event into a major crisis. They include balance sheet issues, such as levels of leverage and liquidity; constraints on the way that institutions behave that are imposed either by regulators or the institutions themselves; and the way in which market participants react to each other in times of both relative calm and stress.
For example, it has been estimated that the losses in the US subprime mortgage market, which triggered the Global Crisis, were roughly equivalent to an equities market fall of 2%. The vast majority of equities market losses of 2% do not lead to any major instabilities, while the subprime losses led to huge downwards spirals given the distribution of losses, the balance sheets, liquidity holdings and interconnections of the market participants, and the lack of common knowledge about who held what and how they were interconnected.
Both regulators and institutions place constraints on how market participants are supposed to behave. Those rules are often motivated by ‘microprudential’ or internal considerations, such as moral hazard or adverse selection. For example, there is a wide range of constraints on the amount of risk an institution can hold, how it can refinance itself and on the collateral it must hold.
Unfortunately, in the spirit of the fallacy of composition, while these rules are meant to guarantee sound microprudential behaviour, they can create dangerous positive (reinforcing) feedback loops. One example would be cases where a fall in asset prices triggers an obligation to raise cash by selling more assets, which pushes prices down further.
In the run-up to the Global Crisis, AIG – the US-based international insurance company – became, through its AIG Financials subsidiary, the world’s largest seller of credit protection in the form of credit default swaps, helped by its AAA rating. In the summer of 2008, AIG’s credit rating was downgraded, obliging it to raise capital to post additional collateral at the height of the Crisis. This triggered a vicious feedback loop that ultimately led AIG to seek a bailout by the Fed.
Analyses based on the idea of endogenous risk show how small shocks like this can snowball into extreme outcomes, which are more destructive than is warranted by the fundamentals of the problem (a phenomenon known as ‘overshooting’). This happens purely because of reinforcing feedback loops originating within the system, without the need for extreme exogenous shocks, provided latent imbalances have been allowed to build up. The outcome for risk in the system as a whole can therefore be fundamentally different from that resulting from the risk management decisions of individual institutions.
We have seen that feedback loops are directly affected by the nature of the regulatory policy environment, which can encourage ‘pro-cyclicality’, a process that is positively correlated with the economic cycle. Bank capital and leverage are two examples of a pro-cyclical process in which risks builds up during stable periods. Banks tend to have surplus capital when the economy is booming, while capital levels drop during recessions. Likewise, economic agents have a tendency to borrow too much during good times and borrow too little in downturns.
Figure 2. Amplification mechanisms lead to pro-cyclicality
As Figure 2 illustrates, pro-cyclicality is often created by the various amplification mechanisms built into the financial system, and is encouraged by risk-weighted capital, mark-to-market accounting and the fact that the strength of financial regulations tends to erode in boom times and come back with a vengeance during and right after crises. Amplifying pro-cyclical feedback loops can also comprise loss and margin spirals, in which fire sales destroy capital and increase risk, which in turn forces further sales, closing the loop.
These feedback loops can operate both in the build-up phase of a crisis (years 1-4) and in the crisis phase (years 4-6). In practice, the loops tend to build up slowly over long periods and to accelerate when reversing in a crisis.
Policy initiatives to reduce systemic risk
To deliver their objectives of maintaining financial stability and preventing and mitigating the impact of financial crises, policymakers in central banks, regulatory authorities and finance ministries strive to design laws, rules, regulations and other devices, especially in reaction to a costly event. Since the Global Crisis, they have launched a large number of policy initiatives, some of which may yield real benefits in terms of controlling the build-up of systemic risk.
Others, however, may perversely increase systemic risk. Laws, rules and regulations that were drawn up – ostensibly to bolster financial stability and limit the build-up of risk – can often become a channel for amplification mechanisms that have precisely the opposite effect. This occurs when multiple rules have inconsistent objectives and interact in unpredicted ways. Often these perverse consequences can remain hidden until it is too late, and this can occur in particular when policies are drafted hastily in response to a crisis.
Policymakers operate under specific mandates and with incomplete information. The new financial regulations they design can often contain embedded ‘constraints’ and devices that can coordinate actions of otherwise seemingly unconnected agents. Such rules often look to be a sensible step in the right direction, but which lead to unintended consequences in more stressed situations and contribute to amplification mechanisms once all indirect effects are taken into account.
Such indirect effects can lead to feedback loops, undesirable coordination and a lack of diversity within the market. Designing a robust regulatory regime is difficult because it is hard to predict how participants’ individual motives will come together to produce overall behaviour in the market. In particular, so-called ‘representative agent’ models used widely by economists are by design unable to guide policymakers in this regard.
Unfortunately, many policy rules may be effective in preventing some types of risk while at the same time creating new, perhaps hidden, sources of risk. These include capital regulations that enable financial institutions to report healthy capital levels while actually holding excessively small and decreasing levels of effective capital.
They also include poorly conceived banking regulations that can lead to the emergence of a ‘parallel banking system’ that itself can come with its own vulnerabilities – what former Fed chairman Ben Bernanke in 2012 called the ‘diverse set of institutions and markets that, collectively, carry out traditional banking functions – but do so outside, or in ways only loosely linked to, the traditional system of regulated depository institutions’.
None of this implies that no remedial actions ought to be undertaken for fear of the potential unintended consequences. Rather, it suggests that research into policy effects ought to incorporate – even if only tentatively – the system as a whole, with all its imperfections, to capture its indirect and feedback effects and minimise the occurrence of unintended consequences. Even the ambiguity under which policymakers operate – the fact of not being sure which model is the right one – can be coherently modelled and taken into account in the first place.
In the research programme we are leading at the LSE Systemic Risk Centre, we aim to develop a set of tools for policymakers to adjust regulations to achieve the twin goals of ensuring the efficiency of the financial system and mitigating the incidence and severity of financial crises. While it is not possible to eliminate systemic risk or the incidence of crises entirely, the objective should be a more resilient financial system that is less prone to disastrous crises while still delivering benefits for the wider economy and society.
Bernanke, B (2012) “Some Reflections on the Crisis and the Policy Response”, Fed chairman’s speech at the Russell Sage Foundation and Century Foundation Conference on ‘Rethinking Finance’, New York.
Danielsson, J (2013), Global Financial Systems: Stability and Risk, Pearson.
Friedman, M (1980), “Video recording: Money and Inflation”, Harcourt Brace Jovanovich, made by WQLN.
Keynes, J M (1936), The General Theory of Employment, Interest and Money.
Minsky, H (1982), Can ‘It’ Happen Again? Essays on Instability and Finance, M E Sharpe.
Systemic Risk Centre (2015), ‘Systemic Risk: What research tells us and what we need to find out”, available at http://www.systemicrisk.ac.uk/.
Zigrand, J-P (2014), “Systems and Systemic Risk in Finance and Economics”, SRC Special Paper No. 1.