Finding itself suddenly short of power, Ohio drew two gigawatts from the neighbouring state of Michigan. In response, transmission lines in both states tripped out. The network was soon overwhelmed by huge fluctuations, caused by cuts in demand in some areas due to blackouts, and surges of power from other stations. Local grids separated from each other to try to control the damage, but to no avail. Failures cascaded through the system, and by the end 256 power plants were offline, and over 50 million people were without power.
Like the electricity grid, the banking system is a vital utility that we all rely on. It is also a huge connected network that controls the flow of money rather than electrons. When one bank or financial inst.i.tution fails, it puts other nodes in the network under increased stress. Not only must they make up the financial slack, but they also come under increased scrutiny themselves. A run on one bank makes customers at other banks twitchy; and sagging lines of credit may cause a fire.
The credit crunch was like a power outage rolling around the world in slow motion. The first inst.i.tutions to go offline were over-leveraged lenders like Northern Rock in the UK and Countrywide in the US. In March 2008 the investment bank Bear Stearns, on the verge of collapse, was taken over by JPMorgan Chase. During the course of the crisis, household names like Merrill Lynch, Fannie Mae, Freddie Mac, Lehman Brothers, and the giant insurer AIG all failed, were taken over under duress, or were rescued by the government.
The credit blackout did not respect national borders: entire countries, such as Iceland, found themselves in the dark, their bank supplies cut off. Some of the worst affected were Eastern European nations that relied on financing from Western banks, such as Hungary, Lithuania, and Latvia. The latter saw an annual house price decline of almost 60 per cent in the year following the crisis.11 In the Middle East, Dubai experienced a similar decline in real estate prices, causing the company Dubai World to freeze payments on its debt in late 2009, with indirect effects on international markets.12 So is there anything we can do to protect ourselves from such failures - or will we always be vulnerable to electrical storms?
The science of networks.
The banking and electrical systems are two examples of technological networks. Others are the transportation network, the telecommunications network, and the world wide web. Similar networks are ubiquitous in nature: biological systems are characterised by complex networks of interacting genes and proteins, ecosystems by predator-prey relationships. And sociologists use social networks to investigate the transmission of ideas and trends through society.
Researchers in the field of network science view such systems in terms of nodes, which represent individuals or agents in the network, and links, which join the nodes and represent interactions of some kind. In a biological model the nodes could represent proteins or cells; in an ecosystem model they could represent species; in a social network they could represent people; in a model of the electrical grid they could represent power stations or consumers; in a model of the economy they could represent firms. For example, one paper published by Domenico Delli Gatti from the Catholic University of Milan and colleagues in June 2008 observed that: "The complex pattern of credit relationships is a natural research issue to be dealt with by means of network a.n.a.lysis. It is straightforward to think of agents as nodes and of debt contracts as links in a credit network ... the default of one agent can bring about an avalanche of bankruptcies [their italics]."13 If the authors had delayed publication a few months, they could have used Lehman"s as an example.
Researchers have found that such networks - be they technological, biological, ecological, social, or economic - often have much in common, and can be divided into certain categories. One is the small-world network, where the connections between individual nodes are arranged in such a way that it takes only a small number of steps to link one node to another. The world wide web has this property, and search companies such as Google exploit it to derive their algorithms. Another category is scale-free networks. The term "scale-free" means that there is no typical or expected number of connections for any node: most nodes have few connections to other nodes, but a small number of hubs are highly connected. An example is the air traffic network: some airports such as Heathrow are global hubs, while smaller regional airports may fly to only a few destinations.
Artificial networks with these and other properties can easily be produced and studied on the computer. Network modelling of the economy has become an active research area, in academia and inst.i.tutions including the Bank of England. One of the key questions that engineers and network scientists are concerned with is network robustness, which often depends strongly on the way in which the network is arranged. Much can be learned from natural systems, such as ecosystems or biological systems, simply because they have been around for a long time so have presumably learned a trick or two. Some "design principles" shared by robust networks - but not currently by our financial system - include modularity, redundancy, diversity, and a process for controlled shut-down.14 Together they provide clues on how we can reduce the chance of another disaster.
Modularity A network"s modularity refers to its degree of compartmentalisation. In, for example, a small-world network, each node is connected to any other node by a small number of connections. This is good if the aim is communication, but in other cases it can be a problem. Scientists have studied the spread of epidemics using detailed network models of artificial societies in which nodes represent individuals, and connections between nodes represent the potential spread of the disease from one person to another. It turns out that one of the main factors determining the rate of spread is the transportation network - the 2009 swine flu pandemic spread so quickly because of long-distance connections through air travel.
The banking system too has become increasingly integrated, and therefore vulnerable to contagion of a different sort. After the Great Depression, the Gla.s.s-Steagall Act was introduced in the US to separate commercial banks, responsible for day-to-day consumer banking, from investment banks, which were primarily involved in speculation. The repeal of this act in 1999 by the Gramm-Leach-Bliley Act dissolved the wall, and allowed banks like Citigroup to go nuts with derivatives, lose billions, and get rescued by the US government. (The same act also led to deregulation of electricity markets and the Enron saga.) On an international level, the degree of financial connectivity between major markets has increased dramatically in recent decades - meaning that if one catches a cold, they all get it.15 Complex living organisms, or natural systems such as food webs, tend to be built up of smaller, weakly connected sub-networks, which reduces the probability of contagion from one area to another.16 The overall topology or structure of the network architecture is also important. A common motif in biological and engineering networks is the "bow-tie" structure, in which multiple inputs (one side of the bow) feed into a central control unit (the knot) to produce multiple outputs (the other side of the bow). An example again is the internet, where a wide variety of material such as web pages, emails, video, and so on, is first compressed into a h.o.m.ogeneous, standardised computer language before expanding again as output on a user"s screen. According to control theorists, who study the control of dynamical systems in engineering, the bow-tie structure has evolved in both natural and man-made systems because it allows a balance between robustness and efficiency.17 The system is quite efficient, because it uses a standardised language to handle all the diverse inputs and outputs, but at the same time it is easy to monitor events and correct mistakes. In finance, the equivalent to a central control module would be a central clearing house for instruments such as derivatives. These are currently often sold over-the-counter, which makes it impossible to measure or control systemic risk.
Redundancy Another trick that nature employs to improve robustness is keeping something in place for backup. If one node or link in the network fails, another can take its place. That extra kidney might seem a waste to carry around until your other kidney fails (or you need to donate one). In financial terms, this supports the idea that banks should retain a higher minimum level of cash reserves, which could be adjusted up for large inst.i.tutions or investment strategies that pose systemic risk.
Much of the appeal of the complex financial products developed in the last decade is that they enabled financial inst.i.tutions to get around reserve requirements. Investment banks such as Lehman Brothers were leveraged at extremely high ratios (over 30 to 1), so they were essentially gambling with other people"s money. The danger, as chairman of the US Federal Deposit Insurance Corporation Sheila Bair told a conference in June 2007, is that "Without proper capital regulation, banks can operate in the marketplace with little or no capital. And governments and deposit insurers end up holding the bag, bearing much of the risk and cost of failure....The final bill for inadequate capital regulation can be very heavy. In short, regulators can"t leave capital decisions totally to the banks. We wouldn"t be doing our jobs or serving the public interest if we did."18 Canadian banks survived the credit crunch relatively unscathed, in large part because they have tougher lending requirements than their American counterparts.19 Diversity A degree of diversity in a system can help it adapt to change. In an ecosystem this equates to a range of species; in the financial system it equates to diversity of trading strategies. On the surface, our financial system would appear to be highly diverse. However, one surprise to come out of the crisis was that everyone appeared to be employing the same strategies. Even adventurous hedge funds, which are supposed to come up with innovative ways to make money, were susceptible to group-think. Intense compet.i.tion between inst.i.tutions meant they were afraid of under-performing their peers, so were actually more likely to adopt the same techniques. As one trader put it, they "talk to each other and have many of the same trades. These are people who say, "I see a pattern, and I"ve got to jump on.""20 The trend was exacerbated by the fact that funds often use quant.i.tative rule-based strategies, which are inherently easy to copy. Banks also adopted near-identical risk models, even though they were known to be flawed, exactly because they were widely accepted by the industry. Complexity scientists are starting to monitor these different strategies, and the relationships between them, in the same way that ecologists monitor species in an ecosystem.21 Controlled shut-down When cells in the human body are damaged beyond repair - say after exposure to toxins or radiation - they are usually targeted for a form of controlled death known as apoptosis. In this process, the const.i.tuents of the cell are taken apart and recycled for use elsewhere in the body. In cancer cells, the apoptotic machinery is disabled, and cells at the interior of the tumour become necrotic - they burst, disgorging their contents in a fashion that harms nearby cells.
When Lehman went bankrupt, its death was necrotic rather than apoptotic. In the US alone, it had over a million derivatives transactions outstanding with some 6,500 trading partners. Figuring out the mess will keep hundreds of lawyers employed for years. Banks also often structure themselves in a deliberately labyrinthine manner in order to avoid taxes, which makes them hard to wind up. Proposals for "living wills" for banks are being considered by inst.i.tutions including the UK"s Financial Services Authority.22 To improve the robustness of our financial system it therefore follows that we should increase modularity, redundancy, and diversity, and provide a mechanism for controlled shut-down. This applies not just to banks, but to other industries such as agriculture or retail, which, as discussed later, exhibit many of the same problems. There"s only one problem: none of these measures would be seen as desirable according to orthodox dogma. The reason is again related to the idea of efficient markets.
Fixing the grid.
According to theory, markets are made efficient if each atom (e.g. individual or company) pursues its own self-interest. Here self-interest refers usually to short-term interest, because if a company neglects the short term it will be taken over by compet.i.tors. And what happens after it dies is irrelevant. Economics likes to live only in the present.
Companies, including banks, therefore spend a lot of time worrying about their own short-term risk, but much less on systemic risk.23 Governments and regulatory inst.i.tutions have also generally gone along with the idea that markets are self-regulating (though after the credit crunch, Alan Greenspan admitted that this idea was "a flaw in the model ... that defines how the world works").24 The financial network is therefore allowed to evolve towards a state that appears highly efficient in the short term, but is constantly acc.u.mulating systemic risk.
Introducing modularity, for example by separating speculative activities from ordinary commercial banking activities, or dividing large global banks into clearly defined national components, would probably reduce short-term efficiency, as would building extra slack and capacity into the system, e.g. by increasing the amount of money that banks need to keep on reserve.25 Such measures can therefore be taken only by a strong regulatory agency. Some progress is now being made - there is certainly a desire for reform in the air - but changes will occur only under protest by the banks, which appear to have learned few lessons from the crisis, except that they can rely on taxpayer bail-outs. Indeed with the collapse of many players, the banking industry is more concentrated than it was before the crisis.
It is interesting to ask whether the credit crunch would ever have happened if politicians and risk experts at banks had been trained or educationally shaped in fields like complexity and network theory rather than orthodox economics.26 When the US government took the decision to let Lehman fail in an uncontrolled manner, it seems that the administration was taken aback by the indirect effects. It was like an untrained apprentice engineer wandering into the control room and unplugging the thick cable with the "DO NOT DISCONNECT" sign above it.27 And the result was nearly lights out for the economy. Three days after Lehman"s bankruptcy, on September 18, the Federal Reserve had to intervene to stop an electronic bank run on US money market accounts. As Representative Paul Kanjorski of Pennsylvania explained, they feared that if it were allowed to continue, "$5.5 trillion would have been drawn out of the money market system of the US, which would have collapsed the entire economy of the US, and within 24 hours the world economy would have collapsed. It would have been the end of our economic system and our political system as we know it."28 The next way to revive economics, then, is to educate our cadre of highly-paid "financial engineers" in the principles and codes of real engineering. This includes building in firebreaks and safeguards to help prevent systemic failure, and developing diagnostic tools for the collection and a.n.a.lysis of network data. "At present," notes the Bank of England"s Andrew Haldane, "risk measurement in financial systems is atomistic. Risks are evaluated node by node. In a network, this approach gives little sense of risks to the nodes, much less to the overall system. It risks leaving policymakers navigating in dense fog when a.s.sessing the dynamics of the financial system."29 Disasters and breakdowns will always occur, but the effects can be minimised and procedures put in place to get the system up and running as quickly as possible (the blackout in the US north-east was repaired in most places in under a day). New ideas and tools from mathematical areas like network theory and complexity can help to frame the problems, test and refine hypotheses, explore and communicate solutions, and motivate changes. As shown by the influence of neocla.s.sical economics, models can have a large effect on the design of financial structures.
So far we have seen that economics derives its authority from its use of mathematical equations, and ideas such as atomism, that are core to our tradition of reductionist scientific thought. In the next chapter, we look at another topic beloved of the ancient Greeks and modern economists alike - the notion of stability.
CHAPTER 3.
THE UNSTABLE ECONOMY.
To understand what is going on we need a new paradigm. The currently prevailing paradigm, namely that financial markets tend towards equilibrium, is both false and misleading; our current troubles can be largely attributed to the fact that the international financial system has been developed on the basis of that paradigm.
George Soros (2008).
There is nothing in this world constant but inconstancy.
Jonathan Swift (1707).
Economists are taught that the economy is intrinsically stable - price changes are small and random, so perturbations are rapidly damped out by the "invisible hand" of market forces. This a.s.sumption would be fine, except that it is contradicted by all of financial history. Booms and busts aren"t exceptions, they are the standard course of things. This chapter shows how the a.s.sumption of stability has been a feature of scientific modelling of natural systems since the time of the ancient Greeks - and why we need to better account for the dynamic, unpredictable, and reflexive nature of the economy.
If there are three words that characterise the orthodox view of the economy, they are efficiency, stability, and rationality. Economists seem compelled to convince the rest of us that the market is some kind of magnificent technological machine that automatically allocates resources with mathematical precision and is immune to shakes, wobbles, or outbreaks of craziness or delirium.
In the 19th century, when neocla.s.sical economics was invented, the a.s.sumption of stability was required because without it, it would have been impossible to solve the equations using the mathematics of the time. However, this excuse is no longer relevant, now that we have computers to do the work. It is therefore strange, as J.-P. Bouchaud noted, that "cla.s.sical economics has no framework through which to understand "wild" markets, even though their existence is so obvious to the layman."1 This seems to be a grave omission - especially since the credit crunch put the Eek! back into economics. So why is it that economists cling to the notion of stability? To find the answer, we must again delve into the historical context.
The highest aim of mathematics has always been to find that which is timeless and unchanging in a world that appears to be in a constant state of flux. The beauty of the Pythagorean theorem about right-angled triangles is that it applies not just to some right-angled triangles, or most right-angled triangles, but to all right-angled triangles, now and at any time in the future. Of course, if you actually draw a right-angled triangle, and measure the sides, and compare the sum of squares of the two sides with that of the hypotenuse, then you will not get a perfect match, but that is because the triangle you drew is slightly flawed, and the measurements have a small error. The law applies only to perfect triangles, which like numbers themselves are mathematical abstractions.
Plato generalised this idea with his theory of forms. According to Plato, every object, such as a table or a chair, is an imperfect version of the Table form or the Chair form, which exist in some higher plane of reality. Everyday objects are subject to change, but their forms live forever. We perceive the world using our senses, but forms can be known only through the intellect - and only they can lead to genuine knowledge.
As discussed in Chapter 1, the quest for unchanging laws still drives much of science. Even if the cosmos is in a constant state of flux, the laws that govern it are considered permanent and immutable.2 Mathematics is a way of fixing nature, a kind of photography device to capture the eternal.
In Arthur Koestler"s cla.s.sic history The Sleepwalkers, he criticised Plato for what amounted to a fear of change: "When reality becomes unbearable, the mind must withdraw from it and create a world of artificial perfection. Plato"s world of pure Ideas and Forms, which alone is to be considered as real, whereas the world of nature which we perceive is merely its cheap Woolworth copy, is a flight into delusion." (Woolworths lent poignancy to this quote by itself disappearing into bankruptcy as an early victim of the credit crunch in the UK.)
Efficiency, stability, and rationality.
Like the ancient Greeks, neocla.s.sical economists saw the economy as a "world of artificial perfection" that was governed by order and stability. Three of the key figures who built the foundations of neocla.s.sical theory were William Stanley Jevons (1835-82), Leon Walras (1834-1910), and Vilfredo Pareto (1848-1923). Although they all pursued an abstract vision of placid stability in their work, it is ironic that they were all strong characters who led interesting and not particularly stable lives. My impression is that they would have collectively run screaming from most present-day university economics departments.
We have already met William Stanley Jevons, who first translated utility theory into mathematical form. Jevons was a true polymath, with interests in chemistry, physics (he published two papers on Brownian motion), botany, meteorology (a paper on cloud formation), economics, social policy, and music.
Born in Liverpool in 1835, Jevons attended University College London for two years, but after the collapse of his father"s iron business he was forced for financial reasons to look for work. Offered a position at the new mint in Australia, he moved to Sydney, where he spent five years, continuing his scientific investigations on the side. He then returned to the UK to complete his university degree. While supporting himself as a tutor and lecturer at Owens College, Manchester, he wrote several papers on logic, then in 1865 won a degree of fame with his treatise The Coal Question, which compellingly argued that the country was on the verge of running out of coal. Promoted to professor, he at first spent most of his time working on logic. He was spurred to return to economics after Fleeming Jenkin sent him a copy of a paper including a geometric interpretation of supply and demand (Figure 1) that resembled Jevons" own theories.
His 1871 book Theory of Political Economy drew explicit comparisons between utility theory and physics. Like a physicist working on abstract problems where effects like friction or turbulence are ignored, Jevons a.n.a.lysed only idealised markets, in which each individual makes decisions based on "a pure regard to his own requirements or private interests," the "intentions of exchanging are known to all," and there is "perfectly free compet.i.tion" between partic.i.p.ants. He compared the price mechanism to the motion of a pendulum, which comes to rest at the ideal balance between supply and demand.
In 1876 Jevons moved back to University College London, but he found his professional duties too stressful and time-consuming and resigned after four years. He was plagued throughout his life by poor health, insomnia, and depression (mental health problems ran in his family, and his older brother and closest sister went insane). He died in a swimming accident near Hastings at the age of 46, leaving behind several thousand books, a huge stock of blank writing paper (he was antic.i.p.ating a shortage), and an enduring reputation as one of the most influential economists in history.3
Fixed point.
Next came Leon Walras. Born in France in 1834 to an economist father, he enrolled at the Paris School of Mines. However, he didn"t enjoy engineering, so he tried a number of other careers including romance novelist, journalist, clerk at a railway company, and bank manager. Influenced by his father, he wrote some papers on economics, and in 1870 he was offered a chair in political economy at the Academy of Lausanne in Switzerland. He later said that economics had provided him with "pleasures and joys like those that religion provides to the faithful."4 Walras is best known for his Elements of Pure Economics, which is considered the founding text of equilibrium theory, a major plank of orthodox economics. While Jevons had considered only simplified examples, Walras simulated the workings of markets for multiple goods, where the price of one good (say wheat) could have indirect effects on other goods (e.g. bread). He realised that the economy had to be modelled as an interconnected whole, like a solar system in which each body exerts a gravitational force on the others. To do this, he wrote out a set of equations that modelled the interactions between sellers and buyers for a range of products. Drawing on the mathematician Louis Poinsot"s book elements de Statique (Static Elements), a copy of which he is said to have kept nearby at all times, Walras argued that because the number of unknowns equalled the number of equations, it was possible to solve the equations. He couldn"t actually solve them himself, but in principle a solution should exist.
Walras" ideas made rather little impact during his lifetime, but his reputation has increased steadily ever since. In 1954 the economist J.A. Schumpeter wrote that "Walras is in my opinion the greatest of all economists. His system of economic equilibrium, uniting, as it does, the quality of a "revolutionary" creativeness with the quality of cla.s.sic synthesis, is the only work by an economist that will stand comparison with the achievements of theoretical physics."5 When Walras retired, he was succeeded by his disciple, the Italian economist and sociologist Vilfredo Pareto. Pareto"s father was an exiled Italian aristocrat and civil engineer. Pareto studied engineering in Turin, and came top of his cla.s.s with his thesis on The Fundamental Principles of Equilibrium in Solid Bodies. On graduation he became director first of an Italian railway company, then a steel company in Florence. He was also involved in politics, railing in favour of liberalism and against government regulation.6 After the death of his parents in 1889, the 41-year-old quit his job, married a young Russian girl, and moved to a villa in the country, where he began writing and giving public lectures on economics. The government reacted to his provocative speeches by having him tailed and closing down his lectures when they could. Pareto was unintimidated (he was an expert marksman and swordsman, which probably helped). His work eventually got the notice of Leon Walras - with whom he shared a background in engineering - and through him the position at Lausanne.
In 1906 Pareto published his Manual of Political Economy, which elaborated on Walrasian equilibrium and extended its mathematical base. It also introduced the idea of Pareto optimality, defined as a state in which any change that makes a person better off will reduce the wealth of someone else. Pareto is more famous today, though, for his empirical discovery of the so-called 80-20 rule. He observed that in Italy and other countries, 20 per cent of the people held about 80 per cent of the wealth. Furthermore, wealth followed a scale-free distribution, which as mentioned earlier means that there is no typical degree of wealth: most people have little money, but a few are fabulously rich.
Along with other neocla.s.sical economists, Jevons, Walras, and Pareto laid the stable base on which modern economics could erect its impressive and imposing structures. In the 1960s, the economists Kenneth Arrow and Gerard Debreu rigorously demonstrated the first welfare theorem, which states that under certain conditions, free markets lead to a Pareto optimal outcome. Any change such as government regulation will only detract from this ideal equilibrium. Of course, the theorem makes many a.s.sumptions, including perfect compet.i.tion, perfect knowledge for market partic.i.p.ants, negligible transaction costs, and so on. During the Cold War, the welfare theorem was promoted as mathematical proof that capitalism, and not communism, was the final fixed point of human development.
The market pendulum.
Neocla.s.sical equilibrium theory did not actually a.s.sume that the economy is completely stable. The market is constantly perturbed, for example, by political events, which it quickly adjusts to. There are also effects due to technological growth, and the so-called business cycle. However, their contributions were a.s.sumed to be relatively small and slow-acting, so that for practical purposes they could be ignored when everyday prices were being considered.
Jevons was actually very aware of the business cycle - members of his family had been bankrupted by the "railway boom" crisis of 1847 - and he was one of the first economists to study it in detail. Inspired by his meteorological research, he believed that it was a periodic phenomenon driven by sunspots. Sunspots affect the weather, which affects agriculture, which affects the rest of the economy. Or as he put it: "If the planets govern the sun, and the sun governs the vintages and harvests, and thus the prices of food and raw materials and the state of the money market, it follows that the configurations of the planets may prove to be the remote causes of the greatest commercial disasters."7 The fact that the average business cycle, which he put at 10.5 years, didn"t match perfectly with the sunspot cycle led him into a long argument with astronomers over the quality of their solar observations. (An enduring feature of neocla.s.sical economics is that if the data doesn"t fit the theory, then the data must be wrong.) While Jevons et al. imported the concept of stability from physics and engineering to the new subject of economics, they couldn"t actually prove that the solutions to their equations would be stable. It was taken as a given that free markets would adjust prices to a particular level, which once attained would remain unchanged, apart from small perturbations. In a sense they were forced to make this a.s.sumption, because the mathematical tools available to them were suited only for studying static systems, or at best, systems that varied in a periodic fashion. And as Jevons wrote, it is "much more easy to determine the point at which a pendulum will come to rest" than to compute its general motion.
However, it doesn"t necessarily hold that, even if a system has a theoretical equilibrium, it will actually be attained in practice; or if it is attained, that it will remain stable. For example, a pen balanced vertically on a table might be at equilibrium, but the equilibrium isn"t very stable since the slightest nudge will cause the pen to topple over. Such questions are the subject of nonlinear dynamics, and the related engineering field of control theory, which also have roots that go back to the 19th century.
The governor.
At the same time that economists were painstakingly constructing their theory of a stable economy, physicists were trying to discover what exactly it was that made a system stable in the first place. A leader in this area was the Scottish theoretical physicist and mathematician James Clerk Maxwell (1831-79).
Maxwell was Jevons" elder by four years, and like Jevons died quite young at the age of 48 (of cancer). His greatest achievement was his mathematical formulation of the laws of electromagnetism, known as Maxwell"s equations, which showed that electromagnetic waves propagated through s.p.a.ce at the speed of light. He concluded that electricity, magnetism, and light were just different aspects of the same underlying phenomenon-a discovery that Einstein described as the "most profound and the most fruitful that physics has experienced since the time of Newton."
During his short but amazingly productive academic career, Maxwell also made important contributions to statistical mechanics, astronomy, and engineering, and found time to produce the first colour photograph (of a tartan ribbon, not his holidays). But the aspect of his work that concerns us here is his cla.s.sic 1876 paper, "On Governors," which is still taught in introductory cla.s.ses on control theory.
The t.i.tle referred to mechanical devices bolted onto steam engines to regulate their speed. The invention of these governors played a vital role in the industrial revolution, because they made it possible for an engine to keep running smoothly, even under different mechanical loads. An example of a governor from the period is shown in Figure 6. If the engine is running too fast, then the two metal b.a.l.l.s, which spin around with the motor, are spread apart by centrifugal force. This actuates a lever that slows the speed. Conversely, when the speed is too low, the b.a.l.l.s descend, and the speed picks up.
Figure 6. A centrifugal governor.8 If the governor is correctly designed, with the weight of the b.a.l.l.s and the length of the arms and so on adjusted correctly, then the system keeps the engine at a speed that is roughly constant; any small disturbance is quickly damped out. However, this wasn"t the only possibility. Maxwell cla.s.sified the different responses as follows:1. The disturbance increases.
2. The disturbance diminishes.
3. An oscillation of increasing amplitude.
4. An oscillation of decreasing amplitude.
In the first case, any slight increase in the speed would cause the engine to run out of control. In the second, speed would adjust itself to a stable level, as desired. In the third, the speed would wildly oscillate between slow and fast, until the machine fell apart. In the fourth, the speed would oscillate but eventually stabilise.
Stability was therefore just one of the available options, and the system had to be carefully designed in order to achieve it - making the b.a.l.l.s too heavy, for example, could actually make things worse. Engineers who a.s.sumed stability, without doing the necessary calculations, could end up with costly repair bills.
But if a system as simple as two metal b.a.l.l.s attached to a steam engine could show such a range of behaviour, how could economists a.s.sume that the economy would instantly achieve a state of perfect equilibrium? Wasn"t it possible that it too would run out of control, or oscillate violently until it was shaken apart, like an unregulated steam engine?
Harmony of tensions.
The reason that steam engines with governors show such complex behaviour, Maxwell found, is because of the presence of feedback loops. A positive feedback loop is one in which a small disturbance is amplified. The cla.s.sic example is a microphone pointed at a speaker; any sound picked up by the microphone gets sent to the speaker, which sends it back to the microphone, which sends the amplified signal back to the speaker, and so on. A negative feedback is one, like the governor, that tends to resist change.
In Maxwell"s a.n.a.lysis of the governor/steam engine system, cases (2) and (4) represent situations in which negative feedback has the upper hand: a change from the set speed is either diminished directly, or the speed oscillates around the set point but the size of the swing shrinks in magnitude. In cases (1) and (3), positive feedback has the upper hand: any disturbance either just grows in size, or it swings between fast and slow in a highly unstable fashion.
Feedback is a concept that came out of control theory and engineering, but feedback loops are ubiquitous in any kind of organic complex system. In the climate system, for example, clouds have a particularly important and sensitive effect on temperature. If the temperature goes up during the day, then water vapour increases due to evaporation, which increases cloud cover, which cools the atmosphere (negative feedback on temperature). But at night, the cloud cover warms the atmosphere (positive feedback). The dual role of clouds makes them extremely hard to simulate, because a small change in the model can lead to a very different balance between these opposing effects.
Biological systems too are characterised by positive feedback loops that allow a rapid response, coupled with negative feedback that provides control. For example, if you cut your finger, a positive feedback loop triggers the rapid production of proteins to create a blood clot and stop the bleeding. These same reactions are normally tightly controlled, because otherwise they would lead to thrombosis. In fact any living system seems to contain this kind of internal tension, a balance between opposing forces. The Greek philosopher Herac.l.i.tus, who was a contemporary of Pythagoras, said it best: "What is at variance comes to terms with itself-a harmony of opposite tensions, as in the bow or the lyre."
Now, if we a.s.sert, as the neocla.s.sicists did, that the economy is constantly at equilibrium, then that is the same as saying that the dynamics are completely dominated by negative feedback. In fact, according to efficient market theory, any perturbation isn"t just damped out over a period of time - it is removed instantaneously. The main source of negative feedback is a.s.sumed to be the law of supply and demand (Chapter 1). If the price of a good rises too high, then supply will increase and the price will return to equilibrium. If the price falls, then supply will decrease and stability will again be restored.
Another example of negative feedback in the economy is the "law of diminishing returns." If a factory increases its workforce by 20 per cent, it probably won"t get 20 per cent more product out the other end, because the factory will become overcrowded. Similarly, if you have a job, then getting another one will not double your productivity because you won"t be able to work as hard. This can be seen as a kind of stabilising force on the economy, because it limits the amount that any single person or company can do or acquire.
Since academic economists are still teaching undergraduate students that the economy is inherently stable; and since the current economic paradigm a.s.sumes, as George Soros points out, that "financial systems are self-correcting and tend towards equilibrium"; and since, as seen later, the risk models used by banks also implicitly a.s.sume that markets are stable and self-correcting; then one might a.s.sume that any positive feedback loops must be either extremely weak or very hard to find.9 So let"s go for a careful search and see if that"s the case.
The reflexive economy.
One source of positive feedback was discussed in Chapter 1. When prices of an a.s.set like housing or a particular stock are going up, they attract a cla.s.s of investors known as momentum buyers, who see people making money and decide to jump on the bandwagon. That acts as positive feedback by bidding prices up even further. If prices are instead decreasing, then momentum investors will sell the a.s.set, thus again amplifying the change in price. A market dominated by momentum investors would be like Maxwell"s cla.s.s (3), in which the system is subject to increasingly large swings until it loses control altogether.
Another kind of positive feedback is due to the network effects seen in Chapter 2. Investors are not independent, but are in constant communication with one another. When people started making money from UK house price rises in the late 1990s, those who were benefiting didn"t keep it as a closely guarded secret. Word got out, and fast. The media amplified the effect by increasing their coverage the more that prices rose. There were far more property shows on TV in 2007 than there were in 1995. So people who might not have considered themselves momentum buyers in 1995 ended up changing their positions.