Share with your friends


Analytics Magazine

Last Word: The Network is the Risk

Spring 2009

By Manmohan S. Sodhi

Recall Sun Microsystem’s advertisement about the “Network is the Computer.” Given all the economic news — no news is good news and therefore so much news is bad news — and given the connectivity in the modern global economy, I am tempted to say that the network is the risk.

Here are some recent headlines and quotes from the Financial Times about the United Kingdom economy. Having just been across the pond and having seen daily headlines like “[Company name] to cut [number] jobs,” I can say there is similar news for the United States and other economies as well:

  • “U.K. debt fears fuel plunge in sterling”
  • “The Bank of England cuts interest rates to the lowest level in its 315-year history”
  • “Chancellor warns … abandon the forecast that the recovery would start in the second half of 2009?
  • “What started with the credit crunch has become a global economic slump, the like of which we haven’t seen for about two decades,” CFO, British Airways
  • “House prices: Average decline of 15.9% for 2008 to a record low for more than 50 years”
  • “New car registrations fall 21.2% in December 2008 compared with December 2007?
  • “U.K. job appointments and vacancies last month plunged to their lowest level for more than 11 years”
  • “Over 1 million claim jobless benefit”

Modern economic networks are complex. Even individual companies’ supply chains are complex with manufacturing in China, financing in London, inventory owned but not located in Switzerland, and IT and data in India. Economics suggests that over time countries will specialize and trade with each other based on their comparative advantage.

Add taxes to this “comparative advantage” reasoning. Taxes are ways for the government to raise as much money as possible and to keep voters or donors happy. It is hard to take away something already in place, so tax codes keep getting more complicated over time as one layer of new taxes or temporary breaks covers another layer. As a result, businesses have to look for how to decrease their tax burden worldwide. It tickles me to see British tourists in Florida and Las Vegas lining up to buy British liquor in duty-free shops (I line up religiously too). Sometimes the queues are so long that you might think the shops are selling liquor for free, not just free of sin taxes.

We have complex networks and complex transactions based on cost efficiencies, at least from an individual or a company viewpoint. All that liquor sloshing back and forth across the ocean can’t be good for the environment or the airlines, but it is cheaper for individuals. But do companies understand the risks? Do the governments of countries understand the risks owing to this complexity?

We could say that operations research (O.R.), or at least seeking cost efficiency, is part of the problem. We can find the optimal solutions in any trade network, at least when asked, and find increasingly complicated solutions in increasingly complicated networks. But cost-efficient solutions can also be brittle; for instance, single sourcing may provide the best cost, but it also increases supply chain risk owing to potential problems with the supplier.

But O.R. can be part of the solution as well. We have the tools to model networks, and we can use these tools to understand risk as well. For instance, simulation could be used to some advantage. I was brought up in graduate school with the orthodoxy that simulation is a tool of last resort. I can see the wisdom of that, but I don’t think we should always be rigid about it. I have seen simulation being used poorly or even misused in practitioner presentations in the financial district in London, so I am not a big advocate for using simulation. If the purpose of modeling is insight, not numbers, then simulation can lead to more numbers than insight. But if we don’t have good tools to model the dynamism (and therefore the risk) in networks, then simulation is something we should look into carefully.

Besides the familiar Monte Carlo simulation, there is agent-based simulation with heterogeneous agents providing the dynamism we seek to explain. There is also complexity modeling, e.g., cellular automata advocated evocatively/powerfully by Wolfram in his book, “A New Kind of Science” (2002); see also Then there is system dynamics, which is not mainstream operations research but could be valuable in capturing the dynamics related to risk.

Moreover, we have mainstream tools for network modeling, as in network flows. We also have graph-theoretic tools like influence networks that we could use.

So we have all the tools, but what is the problem? The problem is modeling to “explain” risk in networks. The limitation may be that we are used to having normally distributed inputs and getting normally distributed outputs, whereas the kind of problems we are experiencing with the economy right now seem to be steady growth followed by unanticipated collapse.

Still, we need to do something. When there is an elephant in the room, we need to talk about it. Our students, our colleagues, even our families will expect us to say something. We could talk about networks and the risks therein.

ManMohan S. Sodhi ( heads the Operations Research group at Cass Business School in London.

Analytics welcomes feedback, editorial ideas and contributed columns and articles. Contact Peter Horner, editor, at:



Report: Automotive industry to invest $3.3 billion in big data in 2018

Big data investments in the automotive industry are expected to surpass $3.3 billion in 2018, according to a report by SNS Telecom & IT. Amid the proliferation of real-time and historical data from sources such as connected devices, web, social media, sensors, log files and transactional applications, big data is rapidly gaining traction from a diverse range of vertical sectors. Read more →

Digital vs. print books: playing favorites can hurt overall book sales

Since the first e-book platform launched in 2007, e-book sales grew to comprise 20 percent of all book sales by 2015. To ensure the increasing popularity of e-books do not undermine the success of their printed counterparts, publishers frequently delay the digital publication date for several weeks after the print edition has been released. Read more →



INFORMS Annual Meeting
Nov. 4-7, 2018, Phoenix


Making Data Science Pay
July 30-31, 12:30 p.m.-5 p.m.

Predictive Analytics: Failure to Launch Webinar
Aug. 18, 11 a.m.

Applied AI & Machine Learning | Comprehensive
Sept. 10-13, 17-20 and 24-25

Advancing the Analytics-Driven Organization
Sept. 17-20, 12-5 p.m. LIVE Online

The Analytics Clinic: Ensemble Models: Worth the Gains?
Sept. 20, 11 a.m. -12:30 p.m.


CAP® Exam computer-based testing sites are available in 700 locations worldwide. Take the exam close to home and on your schedule:

For more information, go to