Systemic Risk Rising

Last night three newsworthy events took place, any one of which could make for an interesting blog entry about risk management: President Obama unveiled another attempt to reverse the economic slump, a credible (but uncorroborated) threat against NYC and Washington DC near the 10th anniversary of 9/11 was announced, and California, Arizona and Mexico had a massive blackout.  It’s this last event that inspired this blog entry.

At a Santa Fe Institute presentation last year, I got to hear Duncan Watts (Yahoo!’s Chief Scientist) talk about risk management at the micro-level (think individual fund) and at the macro-level (think the entire economy).  Duncan Watts is a researcher and author on the topic of networks and complexity dynamics, with titles like “Everything is Obvious: *Once You Know the Answer” and “Six Degrees: The Science of a Connected Age.”  The topic of the presentation, though, was about how, in the right kinds of networks, the process of reducing local risk can in fact increase global risk.

He considered three different types of networks: electrical grids, human contamination (the ‘pandemic’ network), and financial networks.  His Ph.D. thesis was written on the dynamics of electrical networks, and he concentrated on blackouts in the Rockies and Western US.  He talked about how in the 1970’s and earlier, local blackouts were rather common.  A good thunderstorm knocked out your local transformer or took out the power station in the town next door, and you lost power for several hours or a day.  I remember those quite clearly: it was always fun, as a child in the 1970’s , to take out the flashlight and help mom find the candles.  It happened often enough that we knew what to do as soon as the power went out: take out the candles and the cookies.  Sometime in the late 1970’s and early 1980’s, the US electrical system started addressing the problem of local risk by connecting those local power stations and giving one station the ability to pull power, if needed, from different parts of the “grid.”  We reduced the number of these localized blackouts by simply connecting all those power stations so each station had a backup or several backups.  That’s the drop in local risk.  Towns across the country have far fewer blackouts.

2003 Blackout on the East Coast

In 2003, the East Coast of the US and Canada suffered a massive blackout shown in this satellite image.

But something new started happening that never happened before: massive, large-scale blackouts.  Entire states or regions can now lose power because the network becomes “more rigid” and is prone to collapse because of critical single points of failure.  Each one of these blackouts is ultimately caused by one point of failure: a critical device someplace (usually far away from the population that’s affected) is taken off-line, either accidentally or intentionally, and it results in millions of people losing power.  That’s the increase in global risk.  Duncan explained that his research seems to indicate that if all of the local components of a network adopt the same risk management practice, in this case connecting to other nodes in the grid, then that very act makes the network more brittle and more susceptible to global problems, like regional blackouts.

The implication for risk management in financial services, he claimed, was that if every fund and bank starts using the same methodology for their (local) risk management practices, like VaR or Stress Testing, then the entire economy would be more susceptible to global crashes.  The idea is that if everyone is protected against the same thing, then they’re even more exposed to something that can affect all of them.  He was quick to point out that it’s only true for certain types of networks – and energy and financial networks seem to be of that type.  It clearly doesn’t apply to pandemics/health – vaccinating everyone doesn’t increase the global risk (in fact, it greatly reduces the risk).  But if financial networks fall into this category and this claim is true, it puts into question the very efforts that regulators around the world are working towards: a standardization of risk management practices.  Something to think about the next time you find yourself in the dark.

4 Responses to Systemic Risk Rising

  1. Pingback: Systemic Risk Rising | Revolusionline

  2. Orpheus Mall says:

    I believe this post ties in quite well to your previous post on the Swiss franc. Indeed when the Euro was being adopted everyone was touting the advantage of a currency that had the power of an economy of scale. And indeed it seems centralizing the european currency left it vulnerable and local risk was lowered at the expense of global risk. Perhaps an agent based model would have shed more light on this possibility.

    • Orpheus — agreed. The ‘global’ risk across the eurozone seems to be much higher because of the same apparatus that reduced local risk, for some. But in this case, it may be a zero-sum game: Greece’s risk is lowered while Germany’s is increased. What I mean is that if Germany ends up bailing out Greece, then one country’s local risk [Greece’s] is lowered at the expense of another country [Germany]. This is why Germany is so opposed to the ECB issuing Eurobonds for the entire eurozone instead of each individual country taking its own liability. And yes, I think agent based modeling would certainly help in this case.

  3. roman voronka says:

    While it is true that herd mentality when almost all attempt to reduce risk using the same algorithms increases global risks, does this herd behavior also increse the local risk? For some it may but what about the rest?

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 )

Google photo

You are commenting using your Google 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: