by David L. Levy
Managing crises in complex systems
Just when the world was beginning to wake up to the climate change crisis, with a flood of new evidence on the accelerating meltdown of glaciers and polar ice caps, the financial crisis struck. Paul Gilding has termed this convergence of twin crises “The Great Disruption.” At first glance, the twin meltdowns, financial and climatic, appear to be very different problems requiring very different responses. The financial crisis presents as a relatively short-term crisis that resulted from a combination of housing and financial asset bubbles, poor regulation, excessive leverage, and distorted incentives. The climate crisis, by contrast, is driven by the seemingly inexorable rise in human emissions of greenhouse gases, a longer-term problem with a more defined physical structure. The financial crisis calls for fiscal and monetary policy responses combined with new financial regulation. The climate crisis is spawning a wave of regulatory and market responses, from caps on emissions to subsidies for low-carbon energy.
Despite their apparent differences, however, the climate and financial crises are both manifestations of a common problem – the difficulty in predicting and controlling complex dynamic systems riddled with feedback loops, time lags, and non-linear relationships. Though it’s tempting to point the finger of blame (and there is plenty of blame to go around), both systems are susceptible to fundamental, structural problems of governance. Most notably, we rarely recognize that we are careening over a cliff until we are well past the precipice. Moreover, once we recognize that we are spiraling down in a vicious circle, recovery is far from easy.
Complex systems, such as the climate and the economy, have many interconnected parts that interact in intricate and sometimes poorly understood ways (for a primer on complexity, download my book chapter here). They tend to be unstable and unpredictable, though they also exhibit patterned, cyclical behavior. The current recession, while more severe than most since the 1930s, is following an oft repeated trajectory of boom and bust, of overshoot and collapse in asset prices. In a similar way, but on a much longer time scale, the earth’s climate has cycled from ice ages to relatively warm inter-glacials about every one-hundred thousand years. The search for historical parallels can be useful due to this regularity. As Mark Twain famously said, history does not repeat itself, but it rhymes. Systems move in familiar patterns, but never exactly retrace the same path. A host of complex non-linear interactions mean that small differences in starting conditions can lead to very different outcomes. Though hurricanes take familiar tracks in the North Atlantic, we can never be quite sure how strong one might grow or what city it might hit. The current recession could turn out to be a brief dip or herald the onset of second major depression. Occasionally, complex systems can exhibit dramatic collapse, a result of self-reinforcing feedback loops that transform a minor event into a systemic crisis. We are used to the “normal” business cycle where output, investment, employment, asset prices, and confidence interact to create growth or recession dynamics. In the current recession, however, the sudden recognition that vast pools of complex derivatives based on risky mortgage and credit card debt might be worthless led to panic, frozen credit markets, and bank insolvencies. In a similar way, many scientists fear that human emissions of greenhouse gases could lead to a runaway warming effect due to several mechanisms. For example, polar regions tend to warm more rapidly, leading to loss of polar ice and more absorption of solar radiation. Moreover, warming is likely to lead to faster release of carbon dioxide from forests and vegetation. The vast Russian and Canadian tundra regions are thawing, while recent studies demonstrate that the Amazon will become significantly drier, accelerating the shrinking of the rain forest.
In the June McKinsey Quarterly, Michele Zanini reminds us that the severity of these unpredictable disasters, from earthquakes to stock market crashes, actually follows a predictable power law; there is a clear statistical relationship between the frequency and magnitude of the collapse. The so-called butterfly effect is not always at work; sometimes small events may cascade into a major storm, but more typically self-balancing feedback loops help to restabilize the system. Hurricanes dissipate thermal energy, the fuel for storms. Recessions lower interest rates and prices, stimulating investment and consumption. Occasionally, however, the changes that cascade through a system can push it past a tipping point and lock it into a new “phase state”, with a markedly different set of patterns of cycles. As Keynes famously argued, the economy was stuck in a deflationary cycle in the Great Depression and would not naturally bounce back as wages and prices fell. In a similar way, the climate can transition into ice ages lasting tens of thousands of years, in which ice sheets miles thick covered the British Isles and large portions of North America, or hot periods such as the Jurassic era, which lasted millions of years.
It is, then, not surprising that complex dynamic systems are inherently difficult to manage. Controlling the economy has been likened to attempting to steer a car by looking in the rear-view mirror. We can only see where we have been, not where we are heading. And the mirror is often foggy. Data are available after a time lag and may be distorted or offer conflicting interpretations. While the warning signs might seem obvious with hindsight, distinguishing the signal of a financial crisis from the noise of usual variability is difficult in the early stages. Moreover, the recognition of crisis conditions necessitating extraordinary policy measures is a time-consuming social and political process. Nick Kristoff has argued that evolution has predisposed us to worry about immediate tangible threats from tigers and snakes rather than longer term concerns based on analysis and calculation. As a result, prescient prophets of doom are destined to be ignored and we prefer, instead, to assume that what worked (or appeared to work) yesterday will continue to work tomorrow. There are also social and competitive pressures for this. We exhibit herd behavior in markets and crowds, relaxed and optimistic if others are, but vulnerable to panic contagion. As Citigroup chief executive Chuck Prince famously said in mid-2007: “When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing.”
Facing up to the climate crisis is much harder than recognizing the financial crisis. It is unfolding on a time scale of decades rather than months. Unlike the economy, where unemployment, bankruptcies and foreclosures provide more tangible and immediate evidence, the human imprint on the climate is hard to discern for the casual observer. Changes in carbon dioxide levels are invisible, the melting polar ice sheets and tundra are far away, and the oceans are, for now, soaking up much of the excess heat (though we may have reached nature’s overdraft limit).
Even for climate scientists, sophisticated statistical analysis is needed to demonstrate the human impact of greenhouse gases on global temperatures, and we are only just beginning to discern evidence of a link between hurricane intensity and climate change. Moreover, this is our first climate crisis as a sentient civilization. We have experience and institutions to help measure and deal with hurricanes and financial crises, but these are sorely lacking on the climate front.
We do, of course, attempt to peer into the future for both climate and the economy using modeling techniques, but the picture is generally even foggier than that in the rear view mirror. Models are, by definition, simple representations of a far more complex reality. They attempt to capture the major elements of a system and can perform quite well for short-term forecasting under relatively stable conditions. Economic forecasts hold up in reasonable fashion over several months or quarters, as long as there are no sudden shifts in consumer or investor sentiment. Weather forecasts, which rely on very well defined physical relationships between humidity, temperature, air pressure, and other factors, can be quite accurate for three or four days into the future, a bit better than a guess for up to ten days, but have little utility beyond that.
Weather forecasting illustrates an interesting paradox of complex systems; in principle, the weather is a deterministic system based on the well-known physics of atmospheric interactions, so that from a given set of starting conditions, the system ought to unfold in a predictable way. Yet even the most sophisticated weather models that use powerful supercomputers cannot capture the precise initial conditions at every point on earth; they usually work with a spatial resolution of several kilometers and don’t even have accurate data for all these “grid boxes”. Small errors in initial conditions are amplified across the millions of calculations that simulate the dynamic interaction of these atmospheric boxes every few minutes. For climate forecasting, the problems are similar, though different in scale. We are more interested in the average temperature and rainfall in fifty years time at a regional level than whether it will be cold and raining in Boston next Tuesday. But over a timescale of decades, the climate will exhibit more structural shifts with uncertain feedback effects; we don’t know how rapidly polar ice will melt, how fast rainforests will shrink, what might happen to cloud cover, or whether giant ocean currents like the Gulf Stream might slow or even shut down.
Even once the existence of a crisis is widely recognized, the steps required to address it are far from clear and simple. The economy and the climate have huge inertia and respond slowly and somewhat unpredictably to intervention, with the potential for unwelcome side effects. Action on the financial crisis was delayed while debates played out over fiscal versus monetary intervention, and bank nationalization versus recapitalization. Paul Krugman and Niall Ferguson have been recently feuding over the potential impact of large governmental deficits on inflation and interest rates; while Krugman supports aggressive intervention, for Ferguson and the deficit hawks, large scale government deficits will spook the bond markets, raise interest rates, and choke off any recovery. In any event, available policy options are only partial, limited tools; the government does not directly control business investment or consumer confidence. Large parts of the “gray” financial markets, such as hedge funds, non-bank finance, and credit default swaps issued by insurance companies lie beyond the reach of existing regulatory structures.
In a similar way, action on climate has been delayed while various parties argue over the best course of action – cap-and-trade versus carbon taxes versus direct traditional command and control regulation; nuclear power versus renewable energy. Mitigation policies can have unwanted side effects; raising vehicle fuel efficiency lowers the cost of fuel per mile, and so might encourage more driving. Electric vehicles might be charged on coal-intensive power in the US mid-West. Valuable carbon credits help to make air conditioning plants in China appear more profitable, accelerating production of these power hungry appliances. Policy tools tend to use indirect levers, such as the carbon price, which is likely to so low in the US as to have little effect on corporate or consumer behavior (see: Carbon Markets to Serve the Planet). Large sources of emissions lie mostly beyond the reach of current policy; emissions from developing countries, deforestation, international air travel and sea shipping have been out of bounds, till now at least. Moreover, carbon is only one piece of the climate system; we lack the technical tools to directly address other elements, such as melting ice caps, thawing tundra, or ocean currents.
Arguments about how to tackle crises are partly technical, reflecting different understandings of the workings of complex systems. But these differences are also deeply political, reflecting how we think the crisis, and proposals for intervention and change, will affect us in our particular geographic and economic location. It’s easy to claim that “we are all in the same boat”, but the truth is that some of us are in luxury yachts and some are in leaky dinghies. Bankers holding bad loans and homeowners facing foreclosure might want very different forms of government assistance; meanwhile, the majority of the tax-paying and still-employed public are wary of large scale public bail-outs. On climate change, the fiercest proponents of action on climate change are the low-lying countries likely to be swamped by rising sea levels, while the countries who strongly oppose action possess substantial coal and oil reserves. Rich industrialized countries might consider it reasonable to pay 1-2% of GDP to cut their emissions, but developing countries want access to cheap fossil fuels to drive their own industrial transformations.
Managing through a crisis in a complex dynamic system is clearly no easy task. To adapt the steering metaphor, the apt image might be steering a car with fogged up windows, using a greasy steering wheel that’s only loosely coupled to just one of the wheels. Oh, and there are four people in the car arguing about how we got here, where we are heading, and which way to turn the wheel.